Cycling viability of aqueous superconcentrated electrolytes

250
Sorbonne Université Ecole doctorale 397 - Physique et Chimie des Matériaux Chimie du Solide et Energie, Collège de France, UMR 8260 Vers des électrolytes aqueux superconcentrés pour une application dans les batteries Li-ion Par Léa Droguet Thèse de doctorat de Physique et Chimie des Matériaux Dirigée par Jean-Marie Tarascon, Alexis Grimaud et Olivier Fontaine Présentée et soutenue publiquement le 13 Décembre 2021 Devant un jury composé de : Prof. Rosa Palacín Professeure, ICMAB, Barcelona, Spain Rapporteuse Dr. Mathieu Morcrette Ingénieur de Recherche HDR, LRCS, Amiens Rapporteur Dr. Frédéric Kanoufi Directeur de Recherche, ITODYS, Paris Examinateur Dr. Corsin Battaglia Directeur de Recherche, EMPA, Zürich, Suisse Examinateur Prof. Jean-Marie Tarascon Professeur, Chimie du Solide et Energie, Paris Directeur Dr. Alexis Grimaud Chargé de Recherche, Chimie du Solide et Energie, Paris Encadrant Dr. Olivier Fontaine Maitre de Conférence, ICGM, Montpellier Encadrant

Transcript of Cycling viability of aqueous superconcentrated electrolytes

Sorbonne Université

Ecole doctorale 397 - Physique et Chimie des Matériaux

Chimie du Solide et Energie, Collège de France, UMR 8260

Vers des électrolytes aqueux superconcentrés pour une

application dans les batteries Li-ion

Par Léa Droguet

Thèse de doctorat de Physique et Chimie des Matériaux

Dirigée par Jean-Marie Tarascon, Alexis Grimaud et Olivier Fontaine

Présentée et soutenue publiquement le 13 Décembre 2021

Devant un jury composé de :

Prof. Rosa Palacín Professeure, ICMAB, Barcelona, Spain Rapporteuse

Dr. Mathieu Morcrette Ingénieur de Recherche HDR, LRCS, Amiens Rapporteur

Dr. Frédéric Kanoufi Directeur de Recherche, ITODYS, Paris Examinateur

Dr. Corsin Battaglia Directeur de Recherche, EMPA, Zürich, Suisse Examinateur

Prof. Jean-Marie Tarascon Professeur, Chimie du Solide et Energie, Paris Directeur

Dr. Alexis Grimaud Chargé de Recherche, Chimie du Solide et Energie, Paris Encadrant

Dr. Olivier Fontaine Maitre de Conférence, ICGM, Montpellier Encadrant

Sorbonne Université

Ecole doctorale 397 - Physique et Chimie des Matériaux

Chimie du Solide et Energie, Collège de France, UMR 8260

Towards aqueous superconcentrated electrolytes for Li-ion

battery

By Léa Droguet

Ph.D. thesis of Physics and Chemistry of Materials

Supervised by Jean-Marie Tarascon, Alexis Grimaud and Olivier Fontaine

Presented and defended publicly on December 13th, 2021

In front of the Jury:

Prof. Rosa Palacín Professor, ICMAB, Barcelona, Spain Referee

Dr. Mathieu Morcrette Research Engineer, LRCS, Amiens Referee

Dr. Frédéric Kanoufi Research Director, ITODYS, Paris Examiner

Dr. Corsin Battaglia Research Director, EMPA, Zürich, Switzerland Examiner

Prof. Jean-Marie Tarascon Professor, Chimie du Solide et Energie, Paris Ph.D. director

Dr. Alexis Grimaud Research Scientist, Chimie du Solide et Energie, Paris Supervisor

Dr. Olivier Fontaine Associate Professor, ICGM, Montpellier Supervisor

Acknowledgments 5

Acknowledgements

I would like to express my gratitude to my supervisors in Collège de France, Jean-

Marie Tarascon and Alexis Grimaud for giving me the opportunity to work in this

extraordinary environment. Their energy and availability to students are truly inspiring

and their dedication to science is absolutely impressive. I have learnt a lot from working

with and around them. Farther from Collège de France, I would also like to acknowledge

my supervisor Olivier Fontaine.

After completed this three years, I am grateful to Fanny, Thomas, Nicolas, Pierre,

Damien, Romain, Ivette, Benjamin, Charlotte, Jiaqiang and Laura for the great times

spent during this phd journey in and out the lab. Qing, Linje, Biao, Tuncay, Parth and

Anshuman and are also thanks for the calm but always warm office atmosphere. I also

want to thank all the other talented CSE lab members with whom I shared this journey,

all of you ensured a valuable scientific and friendly atmosphere.

I would also like to thank the collaborators without whom part of this thesis would

not have been the same: Marie-Francine Lagadec (CSE, Collège de France, Paris) for the

E-SEM observations, Thomas Marchandier (CSE, Collège de France) for the time spent

on the XRD trials, Maxime Hallot and Christophe Lethien (IEMN, Université de Lille) for

the Al2O3-ALD, Matthieu Courty (LRCS, Amiens) for the DSC experiments and Steven Le

Vot, Marion Maffre and Mathieu Deschanels (ICGM, Montpellier) for their welcome and

help during my stay in Montpellier. I also truly thank Gustavo Hobold, Rui Guo and Betar

Gallant (Department of mechanical engineering, MIT, Cambridge) for the technical

support (Li/LiF samples, XPS and GC-TCD experiments) and the fruitful discussions.

The French National Research Agency through the Labex STORE-EX project (ANR-10-

LABX-76-01) and the Direction Générale de l’Armement through the Agence innovation

defense are acknowledge for the financial support.

I acknowledge Prof. Rosa Palacín and Dr. Mathieu Morcrette for agreeing to review

this thesis. I also thank Dr. Frédéric Kanoufi and Dr. Corsin Battaglia for accepting to be

part of the jury.

Last but not least, I want to express my immensely gratitude to my friends and family

for their support during both joyful and difficult moments. Their support goes well

beyond these three years.

Table of contents 7

Table of contents

ACKNOWLEDGEMENTS ............................................................................................... 5

GENERAL INTRODUCTION AND THESIS OUTLINE ..................................................... 13

CHAPTER 1 –INTRODUCTION TO AQUEOUS SUPERCONCENTRATED ELECTROLYTE AND THEIR USE IN LI-ION BATTERY (LIB) ........................................................................ 19

FROM AQUEOUS SECONDARY BATTERY TO LI-ION BATTERIES (LIB): LOOKING FOR HIGH-ENERGY DEVICES .................................................................................. 20

AQUEOUS SUPERCONCENTRATED ELECTROLYTE: CAN THE MODIFICATION OF THE PHYSICO-CHEMICAL PROPERTIES AND THE INTERFACIAL REACTIVITY UNLOCK THE COMPETITIVENESS OF AQUEOUS LI-ION BATTERIES? .................... 35

FURTHER IMPROVING AQUEOUS SUPERCONCENTRATED-BASED LIB, EXPANDING THE ESW LIMIT AT THE NEGATIVE ELECTRODE SIDE ........................ 48

CONCLUSION OF THE CHAPTER ....................................................................... 58

CHAPTER 2 – CYCLING VIABILITY OF AQUEOUS SUPERCONCENTRATED ELECTROLYTES BASED ON 20 MOL/KG LITFSI AND 20 MOL/KG LITFSI : 8 MOL/KG LIBETI ........................................................................................................................................ 61

INTRODUCTION ................................................................................................ 62

CYCLING PERFORMANCES IN AQUEOUS SUPERCONCENTRATED ELECTROLYTE ON THE NEGATIVE ELECTRODE SIDE: ROLE OF CONCENTRATION, CYCLING RATE AND TEMPERATURE ...................................................................... 65

ORIGIN OF THE PERFORMANCES DECAY: A GAS MONITORING STUDY ........ 71

SELF-DISCHARGE PROTOCOL TO ASSESS AQUEOUS SUPERCONCENTRATED ELECTROLYTES VIABILITY DURING RESTING PERIOD ............................................ 78

CYCLING VIABILITY ON THE POSITIVE SIDE: A GAS MONITORING STUDY..... 87

CONCLUSION OF THE CHAPTER ....................................................................... 91

CHAPTER 3 – INSTABILITY OF NATIVE SEI LEADS TO THE DRYING OUT OF AQUEOUS SUPERCONCENTRATED LI-ION BATTERY ........................................................................ 95

INTRODUCTION ................................................................................................ 96

PROBING THE SEI INSTABILITY IN AQUEOUS SUPERCONCENTRATED ELECTROLYTES ................................................................ 97

8 Table of content

IMPACT OF WATER CONSUMPTION ON ELECTROLYTE CRYSTALLIZATION 106

ACTIVATION ENERGY OF DIRECT AND INDIRECT HER IN WISE ................... 111

DISCUSSION AND CONCLUSION OF THE CHAPTER ........................................ 119

CHAPTER 4 –MIMICKING INORGANIC-BASED SEI WITH LIF-COATING. UNDERSTANDING OF INORGANIC SEI LIMITATIONS IN WATER-IN-SALT ELECTROLYTE. ...................................................................................................................................... 127

INTRODUCTION .............................................................................................. 128

LIF SOLUBILITY LIMIT IN AQUEOUS SUPERCONCENTRATED ELECTROLYTE 129

USING LI/LIF-COATING TO MIMIC INORGANIC-BASED SEI. EXPOSURE TO ATMOSPHERE ENVIRONMENT, AQUEOUS SUPERCONCENTRATED ELECTROLYTE AND COMPARISON WITH THE BEHAVIOR OBSERVED IN ORGANIC ELECTROLYTE ............................................................................................................................. 134

COMPARISON OF LIF BEHAVIOR WITH AL2O3-COATED LI SAMPLE ............. 140

FILLING THE STRUCTURAL DEFECTS BY PRESOAKING IN ORGANIC ELECTROLYTE: ASSESSMENT OF THE IMPORTANCE OF AN ORGANIC-INORGANIC BASED SEI ............................................................................................................ 144

CONCLUSION OF THE CHAPTER ..................................................................... 150

GENERAL CONCLUSION AND PERSPECTIVES .......................................................... 155

REFERENCES ............................................................................................................. 161

APPENDIX ................................................................................................................ 187

MATERIALS & METHODS ......................................................................................... 209

MATERIAL PREPARATION ............................................................................ 210

ELECTROCHEMICAL CHARACTERIZATIONS .................................................. 214

PHYSICO-CHEMICAL CHARACTERIZATIONS ................................................. 221

DATA TREATMENT ....................................................................................... 228

PYTHON ....................................................................................................... 236

LIST OF ABBREVIATIONS ......................................................................................... 237

Table of contents 9

RÉSUMÉ EN FRANÇAIS ............................................................................................ 243

GENERAL INTRODUCTION AND THESIS OUTLINE

12 General introduction and thesis outline

General introduction

The replacement of fossil fuels by renewable energies is at the center of the energy

transition critically needed to limit climate change. However, limitations of greenhouse

gases emission (carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O) and

fluorinated gases) may be in conflict with the energy demand. Indeed, as calculated by

the International energy agency (IEA), the latter is set to increase by 4.1 % in 2021,

mostly due to the increase in demand from emerging markets and developing

economies. Hence, despite the contraction of the demand due to the Covid 19

pandemic, in 2020 energy consumption exceeds by 4 % the 2019 level, going back to

pre-Covid level. Furthermore, regardless of the scenario taken into account, the energy

demand continuously increases (see Figure Introduction. 1 and the description of the

scenario in the caption of Figure Introduction. 1).

Figure Introduction. 1 (a) Forecast of the global energy demand increase indexed to their level in 2019. Scenario envisioned: (i) pre-Covid scenario, (ii) stated policies which correspond to a situation back to normal, i.e. similar to pre-Covid one, in 2022 and (iii) delayed recovery scenario corresponding to a prolonged crisis which would be back to pre-pandemic level in 2025 with strong impact on the energy demand growth). Adapted from Ref1. (b) Forecast of the worldwide CO2 emissions (in giga ton per year) until 2030. Adapted from Ref2.

As a consequence, impactful policies and major investments in clean energy (up to

four trillion USD a year until 2030, as mentioned in the world energy outlook of 2021)

need to be set up to reach the sustainable development scenario that considers the

fulfillment of the objectives discussed during the Paris agreement (which aim to limit

the increase in temperature to 1.5 °C by the end of the century). Even more challenging

is the Net zero emission scenario by 2050 (rather than 2070 in the sustainable

development scenario). Thus, changes in energy production, reduction in energy

consumption, improvement in energy efficiency and innovation in carbon capture can

be seen as the main pillars to succeed in meeting these goals, as illustrated in Figure

General introduction and thesis outline 13

Introduction. 1b. To succeed in this great challenge, electrification can be seen as one

of the main path to develop, among others such as promoting clean energy innovation

or changing societal behaviors. Indeed, 75 % of the world electricity demand should be

produce from low-carbon energy sources by 2030 (compared to less than 40 % in 2019).

Though, electricity produced by renewable energies such as solar or wind power is

intermittent, therefore the urgent need for storage devices.

Many applications from electrical, thermal, and mechanical to electrochemical

devices can store energy. Among electrochemical ones, batteries market is greatly

increasing within the last ten years and is forecasted to grow up even more with the

development of Li-ion batteries (LIB) in the transportation market, leading to a global

market opportunity to 2050 worth USD 16.2 trillion. Indeed, while Lead-acid battery

used to be the most widely used technology, 2020 has seen LIB to dominate the market.

This trend is confirmed by forecasts that announce that the LIB market will almost

double within the next 10 years, as shown in Figure Introduction. 2a. Such conquest of

the market is enabled by technology advances due to cell chemistry innovation, cell

engineering and optimization and growth of manufacturing volume than enable to

improve performances and reliability while decreasing the cost of the cell (see Figure

Introduction. 2b). Though, as LIB battery market is increasing markedly, the anticipation

of LIB battery recycling as well as the development of more sustainable battery

component is of crucial importance.

Figure Introduction. 2 (a) Battery market demand in the world (in $) from 2015 to 2030. (b) Forecast of global annual Li-ion battery (LIB) deployment in all markets in GWh and LIB cell average cost (in purple) as function of time. The cell is based on a 40 Ah pouch cell made of LiNi0.6Mn0.2Co0.2O2 (NMC622) as positive electrode and graphite as negative. Adapted from Ref3,4.

14 General introduction and thesis outline

Outline of the thesis

Therefore, the aim of this thesis is to study the practicability of developing LIB using

aqueous superconcentrated electrolytes, so-called Water-in-salt electrolytes (WiSE).

Indeed, this technology would enable the use of environmental-friendly solvent, i.e.

water, while achieving performances close to commercial LIB.

The thesis is structured in five chapters detailed as follow.

The first chapter briefly summarizes the Lead-acid, Nickel Cadmium, Nickel Metal

hydride and commercial LIB battery technologies. Then, the concept of aqueous

superconcentrated electrolyte is introduced, giving details on how the solvation

structures of such electrolytes impact their physico-chemical properties and interfacial

reactivity and consequently the battery performances. Eventually, the modifications

regarding electrolyte composition proposed in the literature in the last five years are

detailed.

The second chapter describes a systematic study that assesses the practicability of

WiSE-based LIB by decoupling parasitic reactions at the negative and positive electrodes

and by performing electrochemical characterizations during cycling and self-discharge

tests as well as operando gas monitoring. The stability of these electrolyte was

determined to suffer from water reduction at the negative electrode both during cycling

and resting periods, unlike positive electrode where very limited water oxidation was

observed.

Then, in the third chapter, we analyze the stability of the native solid electrochemical

interphase (SEI) by electrochemical characterizations, using cyclic voltammetry and

impedance spectroscopy. Besides, the irreversible consumption of water which leads to

increase in electrolyte concentration was confirmed by differential scanning

calorimetry. Based on the electrochemical results, the rate of water consumption during

resting period was found to be smaller than the one during cycling, though in the same

order of magnitude. The rate of Li+ delithiation was also found to be different when

comparing constant current continuous cycling and self-discharge experiments. To

understand these observations, the activation energy of self-discharge process and

direct water reduction were determined. Eventually, to summarize and rationalize these

electrochemical results, a figure-of-merit was made to compare the performances of

WiSE-based LIB with that of commercial LIB and other commercial aqueous secondary

General introduction and thesis outline 15

batteries with the largest production volume (Lead-acid, Nickel-Cadmium and Nickel

Metal hydride).

Moreover, as LiF is known to be the SEI component in WiSE, and following the first

two chapters in which the instability of the native SEI was highlighted, LiF solubility

measurements were performed in WiSE electrolyte to determine if LiF dissolution is

responsible for the SEI instability. Then, an artificial LiF conformal layer was deposited

onto metallic Li (Li/LiF) to assess the protective-power of inorganic coatings against

water-based electrolytes. Gas chromatography-mass spectrometry was done to analyze

the reactivity of the Li/LiF sample to WiSE exposure as well as the impact of presoaking

Li/LiF samples in organic electrolyte to improve the efficiency of these coating toward

water reduction. Similar observations were made with conformal Al2O3 coatings

prepared by atomic layer deposition (ALD).

Altogether, this manuscript highlights that despite the formation of a LiF-inorganic

SEI in 20 m LiTFSI, water reduction cannot be avoided at the negative electrode and

artificial inorganic coatings are not sufficient to prevent water to access the negative

interface. Therefore, the cathodic challenge remains unsolved.

CHAPTER 1 –INTRODUCTION TO

AQUEOUS SUPERCONCENTRATED

ELECTROLYTE AND THEIR USE IN LI-ION BATTERY

(LIB)

18 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

From aqueous secondary battery to Li-ion batteries (LIB): looking for high-energy devices

Secondary or rechargeable batteries are nowadays widely used in many applications,

from grid storage, electrical transportation to portable devices. As any electrochemical

devices, they are made by assembling two electrodes with different redox potentials,

separated by an electrolyte which is generally composed of a solvent and a supporting

salt. The electrons circulate through the external electrical circuit, thus enabling the

reversible energy exchange with the user, as shown in Figure 1. 1.

The history of secondary batteries started more than 150 years ago, in 1859, by the

development of rechargeable Lead-acid batteries by Gaston Planté who designed the

Pb ǁ H2SO4 ǁ PbO2 cell, as described in Figure 1. 1. The associated electrochemical

process is based on a dissolution/precipitation mechanism directly involving the acidic

electrolyte. Thanks to their low cost, low self-discharge, maturity and reliability, Lead-

acid batteries are still widely used for unit power sources (UPS), starting lighting and

ignition in vehicles (SLI) and emergency lighting. However, the specific energy and

energy density - calculated as function of the cell voltage (V) and the cell capacity (Ah/kg)

(see Equation 1. 1) and expressed either in Wh/kgcell or Wh/Lcell - is limited to 60

Wh/kg5,6.

Equation 1. 1:

𝐸𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐𝑊ℎ/𝑘𝑔

= 𝑄 (𝐴ℎ

𝑘𝑔) ∙ 𝑉 (𝑉)

Following this pioneering work, Nickel-Cadmium (Ni-Cd) batteries were introduced in

1909. For this technology, the negative electrode (Cd) endorses a conversion reaction

while the positive electrode, NiOOH, is reduced or oxidized upon discharge or charge,

respectively, as illustrated in Figure 1. 1b. Besides, their low cost and high rate

performances favor the use of Ni-Cd batteries for power tools, and they were introduced

in early mobile phone model. However, the “memory effect”, the toxicity of Cd and the

limited specific energy (< 60 Wh/kg) limit the overall performances of this chemistry and

called for the development of other technologies. Therefore, from the mid-80s7, a

second generation of rechargeable batteries was introduced with the design of nickel-

metal hydride (Ni-MH) batteries using alkaline electrolyte. Ni-MH batteries use a similar

positive electrode than Ni-Cd, NiOOH, but the electrochemistry at the negative

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 19

electrode relies on intercalation mechanism rather than conversion one (see Figure 1.

1c). Indeed, upon discharge the hydrogen de-intercalates from the metal hydride (MH)

to form a metal alloy (M), and combines with hydroxide anions from the electrolyte to

form water. Ni-MH batteries reach high capacity (110 Wh/kg) with a lesser “memory

effect” than in Ni-Cd ones. Therefore, they were implemented in commonly-used AA

and AAA cells for portable devices, as well as in the first hybrid-electric vehicles (Toyota

Prius). Aside from these three chemistries that, until 2015, represented the commercial

batteries with the largest production volume and USD turnover3,4, several other

aqueous systems were developed in the past decades, including Nickel-iron, Silver-zinc

or Silver-cadmium7 batteries, which will not be discussed in this section.

Figure 1. 1 Chemical reactions taking place in (a) Lead-acid, (b) Nickel-Cadmium (Ni-Cd), (c) Nickel-metal hydride (Ni-MH) batteries.

Despite these advances in cell performances, the electrochemical stability window

(ESW) of water, and thus of aqueous devices, is limited to 1.23 V, as shown by the yellow

zone in Figure 1. 2a. Above this limit, the hydrogen evolution reaction (HER) occurs upon

reduction while the oxygen evolution reaction (OER) happens upon oxidation, as

described by Equation 1. 2 and Equation 1. 3, respectively.

20 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Equation 1. 2: Hydrogen evolution reaction (HER).

2 ∙ 𝐻2𝑂 + 2 ∙ 𝑒− = 𝐻2 + 2 ∙ 𝐻𝑂− 𝑖𝑛 𝑎𝑙𝑘𝑎𝑙𝑖𝑛𝑒 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠 2 ∙ 𝐻+ + 2 ∙ 𝑒− = 𝐻2 𝑖𝑛 𝑎𝑐𝑖𝑑𝑖𝑐 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠

Equation 1. 3: Oxygen evolution reaction (OER).

2 ∙ 𝐻2𝑂 = 𝑂2 + 4 ∙ 𝐻+ + 4 ∙ 𝑒−

Cycling aqueous batteries within the thermodynamically stable potential window of

water to avoid these parasitic reactions drastically limits the battery voltage and thus

the specific energy, as shown in Figure 1. 2b. One obvious way to increase the energy

density is to extend the operating voltage beyond the stability window while finding

means to handle the gas generated during cycling, as implemented in Lead-acid, Ni-Cd

and Ni-MH batteries and discussed in greater details in Chapter 3. To overcome the

voltage limitation in aqueous environment, Li-based batteries were introduced using

organic electrolyte to replace aqueous proton-based chemistries. Lithium was chosen

for its light weight (6.94 g/mol), its low redox potential (ELi+/Li0 = -3.04 V vs ESH) and its

high theoretical capacity (3862 mA.h/g), thus promising high energy batteries, as

illustrated in Figure 1. 2b.

Figure 1. 2 (a) Pourbaix diagram representing the operating voltage of Lead-acid, Ni-Cd and Ni-MH batteries and the electrochemical stability window (ESW) of water (yellow zone) defined by the hydrogen evolution reaction (HER) upon water reduction and the oxygen evolution reaction (OER) upon water oxidation. (b) Ragone plot of secondary aqueous batteries and organic Li-ion batteries (LIB). Adapted from Ref5–9.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 21

The development of Li-based batteries started in the 1970s with the use of metallic

Li as negative electrode and an intercalation electrode as positive one. This technology

was implemented by coupling the use of metallic Li (see Equation 1. 4) with the

discovery, notably by Stanley Whittingham, of the reversible insertion of Li+ cation into

chalcogenide-based materials such as TiS2 or MoS210–13, as described in Equation 1. 5.

These systems were then commercialized by Exxon in 1972 using TiS2 as positive and

Moli Energy in the late 198014 using MoS2 as positive.

Equation 1. 4: Intercalation of Li+ cation in the positive electrode.

𝐿𝑖+ + 𝑒− 𝐶ℎ𝑎𝑟𝑔𝑒→

𝐷𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒←

𝐿𝑖

Equation 1. 5: Reaction at the metallic Li electrode.

𝐿𝑖𝑀𝐴2

𝐶ℎ𝑎𝑟𝑔𝑒→

𝐷𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒←

𝑀𝐴2 + 𝐿𝑖+ + 𝑒−

with M, transition metal and A , oxygen or sulfur

Li metal battery behavior upon discharging and charging is illustrated in Figure 1. 3.

The metallic Li negative electrode is separated from the positive electrode by a

separator soaked with the aprotic organic liquid electrolyte. Aluminum is used as current

collector on the positive side while copper is generally used as current collector on the

negative side since metallic Li is known to alloy with Al, inducing a volume change

causing pulverization of Al. As shown in Figure 1. 3b, during charge, metallic Li is plated

at the negative, which can lead to dendrite growth that can short-circuit the cell and

inflame the organic electrolyte, causing severe safety issues15. One way to improve

safety was to replace the liquid flammable electrolyte by a less flammable polymer one

which also prevents dendritic growth. However, to ensure good conductivity (σ) of Li+

cations, polymer electrolytes need to be heated up to ≈ 60 °C, thus limiting the user-

friendliness for portable applications.

22 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 3 Scheme of (a-b) a Li metal battery, (c) a Li-ion cell. In this schematic view, only Li+ cations are drawn. Solvent molecules and counter anions are not shown. Ionic and electronic flows are symbolized with arrows. (a) Upon discharge. (b) Upon charge. (c) The metallic Li negative electrode is replaced by an insertion electrode to make a Li-ion cell.

In 1980, John Goodenough and coworkers replaced sulfur-based positive electrode

by oxygen-based one (thus taking advantage of the greater electronegativity of oxygen)

and paved the way for the integration of lithium-containing transition metal oxide Li1-

xCoO2 (LCO) positive electrodes with high potential (ELi insertion > 3 V vs Li+/Li). Following

this pioneering work, numerous studies were carried out, including for reducing the Co

content that has rapidly become a concern for large scale commercialization due to

ethical, cost, and toxicity as well as due to limitations in the reversible capacity achieved

by LCO – reversible discharge capacity of 140 mAh/g is attainable in practice, compared

to 275 mAh/g theoretically available. Therefore, many researches then focused on

partially substituting Co in LiMO2 (M= Co, Ni, Mn, Al ...) layered materials. While Co was

found necessary to stabilize the electrode structure16,17, it was partially substituted with

Ni to achieve greater capacity (> 150 mA.h/g). Besides, the partial substitution of Ni by

Al or Mn was found to thermally stabilize the electrode, thus giving birth to the NCA and

NCM (or NMC)-families in which several ratio of Ni:Co:Al or Ni:Co:Mn were

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 23

developed18,19. Cobalt-free positive electrodes also emerged relying on cheap and

abundant materials as well as their thermally stable properties, despite their lower

capacity and potential. Mn is generally used in spinel structure20, LiMn2O4 (LMO) and Fe

with polyanionic compounds ((XO4)3- (X=S, P, Si, As, Mo, W)), with the well-known olivine

LiFePO4 (LFP) for example.

Besides achieving greater potential and reversible capacity, the development of

lithiated positive electrode unlocked the use of non-lithiated intercalation electrode to

replace metallic Li as negative material, giving birth to the so-called Li-ion rocking-chair

battery21, as illustrated in Figure 1. 3c. To realize the concept of high potential rocking-

chair battery, intercalation material operating at low potential were required.

Carbonaceous materials (soft or hard carbons) were first envisioned as alkali-ions

intercalation materials. However, Li+ intercalation in these materials (below 1 V vs

Li+/Li)22,23 was initially found to cause electrode exfoliation, owing to the use of

propylene carbonate (PC)-solvent which co-intercalates during charge, thus initially

impeding the commercialization of LIB.

Indeed, commercial electrolyte solvents for aprotic batteries are generally based on

carbonate esters, as they enable high Li-salt dissolution and dissociation. One cyclic and

one linear carbonate are generally mixed to combine their properties. Linear esters such

as dimethyl carbonate (DMC), ethyl methyl carbonate (EMC) or diethyl carbonate (DEC)

have low viscosity (0.59 mPa.s for DMC, 0.65 mPa.s for EMC and 0.75 mPa.s for DEC),

thus enabling good transport properties. In contrast, cyclic ester such as ethylene

carbonate (EC) or propylene carbonate (PC) show high dielectric constant (89.78 for EC

and 64.92 for PC), and thus preferentially solvate Li+ cations. The key to realize

commercial LIB was the replacement of PC co-solvent by EC, preventing solvent co-

intercalation into graphite while forming a stable passivation layer, and thus improving

the reversibility of Li (de)intercalation into graphite24,25.

Eventually, in 1983, Yoshino was able to successfully cycle LCO with a petroleum coke

negative electrode26. Following this demonstration, LCO/petroleum coke cells were

commercialized by Sony in 1991. Then graphite electrodes were developed without

solvent co-intercalation24 and remains widely used thanks to its high capacity (392

mAh/g at low potential (E< 0.3 V vs Li+/Li)) and good battery performances with high

coulombic efficiency27.

Moreover, safety, low cost, environmental friendliness, wide liquid stability

temperature range and a broad electrochemical stability window are essential

24 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

properties for electrolytes28. Therefore, electrolyte research is tailored by compromises

between all these properties. Moreover, despite the extensive research for enlarging

the ESW, liquid electrolytes are in majority unstable at low and high potentials, i.e. at

potentials at which the negative (< 1 V vs Li+/Li) and positive electrodes (> 4 V vs Li+/Li)

operate. Therefore, the stability of the LIB relies on the formation of a passivating layer

on the electrodes that prevents further electrolyte degradation by being electronically

insulating while preserving Li+ cations conduction. Hence, at the negative

electrode/electrolyte interface, a solid electrolyte interphase (SEI) is formed following

the electrolyte reduction29. The first SEI model was developed in 1979 by Peled et al.30.

Then, Aurbach et al.31 developed a 2D SEI model based on one inorganic inner-layer

close to the electrode surface, containing LiF compounds among others, and a porous

organic outer-layer based on polymeric species. At the positive electrode/electrolyte

interface, a cathode-solid interphase (CEI) was also reported for some

electrode/electrolyte combinations32–34. Besides, these interphases must ideally have a

uniform morphology as well as being able to accommodate volume changes upon Li+

insertion/deinsertion to be efficient.

In order to rationalize the SEI/CEI formation and thus to select the best electrolyte

composition, HOMO/LUMO energy diagrams are widely used. Indeed, a HOMO/LUMO

energy diagram describes the energy level of the highest occupied molecular orbital

(HOMO) and the lowest unoccupied molecular orbital (LUMO) calculated for isolated

molecules, as shown in Figure 1. 4. Thus, one can potentially correlate the energy levels

to oxidation or reduction potentials of the electrolyte that can form the CEI (see yellow

color in Figure 1. 4) or the SEI (see red color in Figure 1. 4), respectively. Following this

framework, the band gap Eg defined by ELUMO-EHOMO would be directly related to the

ESW35. However, the HOMO and LUMO energy levels are calculated for isolated

molecules (solvent molecules or anions). Such calculations do not take into

consideration the interactions between all the molecules contained in the electrolyte

and the solvation structure that will undoubtedly change the electrolyte reactivity, as

discussed in greater details in section 1.3 and 2. Therefore, as mentioned by Peljo et

al.36, the ESW is more accurately defined by the redox potentials of the electrolyte,

which can only be inferred knowing the exact oxidation and reduction reactions and

their associated Gibbs free energies. To illustrate this point (Figure 1. 4 in blue), taking

water as an example, the bang gap of water is calculated to be 8.9 eV37 whereas the

thermodynamic ESW is known to be 1.23 V.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 25

Figure 1. 4 Energy diagram (in eV, on the left) and corresponding potentials (in V vs Li+/Li, on the right) of the solid electrolyte interphase (SEI in red) and the cathode electrolyte interphase (CEI in yellow) formation at the surface of negative and positive electrodes during battery charge. Adapted from Ref35,36.

In commercial LIB, diluted electrolytes are generally employed, with concentrations

close to ≈ 1 M. This “optimal” concentration reflects a compromise between viscosity

(ionic mobility) and conductivity (dissociated charge carrier number) to ensure good

transport properties38,39, as shown in Figure 1. 5a.

26 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 5 (a) Ionic conductivity (blue) and viscosity (red) at 25 °C as function of the electrolyte concentration (1 mol/L (M), 2 M, 3 M, 4 M, 5 M, 6 M and 7 M LiTFSI in DOL:DME (1:1 vol%)). (b) Li+ transference number as function of the electrolyte concentration. Adapted from Ref40.

Nonetheless, upon increase in concentration and when reaching superconcentration

regime (leading to mass and volume of salt greater than the solvent ones), the solvation

structure of ions changes, thus modifying bulk and interfacial electrolyte properties.

Indeed, a competition takes place for Li+ solvation between scarce organic solvent

molecules and anions, both being Lewis bases (i.e. capable of solvating Li+). Therefore,

ionic association switches from solvent-separated ion pairs (SSIP) in diluted electrolytes

to the formation of contact-ion pairs (CIP) or ionic aggregates (AGG) in

superconcentrated ones, as shown in Figure 1. 6.

Figure 1. 6 Solvation structure from diluted to superconcentrated electrolyte. SSIP corresponds to solvent-separated ion pairs, CIP to contact-ion pairs, AGG to aggregate. Image taken from Ref39.

Consequently, the solvation sheath of Li+ cation that is reported to be fourfold

coordinated in diluted electrolytes41,42, was found to be composed of two aprotic

solvent molecules and two anion molecules in superconcentrated electrolytes43.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 27

Moreover, well-dissociated imide-based salts using anions such as TFSI- anion, for

example (see Table A.1. 1 at the end of the manuscript for anion structure), are

preferred for the elaboration of superconcentrated electrolytes as they can be dissolved

in greater amount in all polar solvents, thus reducing the number of free solvent

molecule in the bulk. Regarding the solvent, it must be selected for its strong ability to

solvate Li+ cations.

Looking at bulk properties, in superconcentrated electrolytes, although the overall

conductivity -that is related to the capability of the electrolyte to carry the ionic current-

is lower than in diluted ones, the transference number of Li+ -that considers the

proportion of ionic current taking by each ionic species- can be greater than in diluted

electrolyte, as illustrated in Figure 1. 5b. Indeed, a change in cation motion from

vehicular (related to cation motion with its solvation sheath) to partially structural

(when cations hop via a serial of ion association/dissociation from one anion or solvent

molecule to another) was proposed to explain the increase in transference number39,44.

However, full understanding of ionic transport in superconcentrated electrolyte remains

unclear44. Besides, for superconcentrated electrolytes, the solvent volatility diminishes

and the thermal stability is enhanced44 since the energy needed to desolvate the solvent

molecules becomes greater due to the diminution of free solvent molecules. The later

effect also kinetically prevents electrode and SEI dissolution mechanism to occur45, thus

reducing side reactions. However, the high viscosity of the electrolyte remains a major

drawback as it increases the cell impedance. In addition, from a practical point of view,

the wetting of the electrodes/separators during battery assembly requires a much

longer pre-treatment for this class of electrolytes.

Furthermore, superconcentration modifies the interfacial organization and thus the

reactivity at both the positive and the negative electrodes. At the positive, anions

populate the interface sufficiently to repulse solvent molecules to reach the inner-

Helmholtz layer and thus prevent their oxidation when compared to dilute electrolytes

for which solvent molecules are present in this layer (see Figure 1. 7a and b)46,47.

Moreover, anions are stable enough at high potential not to be oxidized prior to Li+

deintercalation, thus expanding the ESW (≈ 5.2 V vs Li+/Li for TFSI- oxidation48). Besides,

the aforementioned double layer effect coupled with the lower amount of free solvent

molecules reduce the corrosion of the aluminum current collector. Hence, unlike for

diluted imide-based electrolyte that cannot form AlF3 passivating layer that stops the

corrosion, as PF6- does,49 switching to superconcentrated regime allows for using

aluminum current collector. Indeed, the few Al3+ cations formed upon oxidation of the

28 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

current collector will preferentially interact with the anions to form [Al(anion)x]3-x

complexes. Since the viscosity is high, the complexes diffusion through the bulk

electrolyte is reduced stabilizing the interface46.

Figure 1. 7 Snapshots of the interfacial structure positively polarized from molecular dynamic (MD) simulations of (a) diluted electrolyte based on 1 M LiPF6 in EC:DMC50 (b) superconcentrated electrolyte based on LiTFSI(DMC)1.2 51. Adapted from Ref50,51.

At the negative electrode, upon charge, the inner-Helmholtz layer of

superconcentrated electrolyte is compacted and the anions coordinated to Li+ cations

come closer to the electrode surface, as illustrated in Figure 1. 8a and b. Owing to this

specific double layer structure, a direct reduction mechanism of the anions was

proposed following density functional theory (DFT) calculations. Indeed, a shift of the

anions LUMO to greater energies than the solvent LUMO was calculated. Therefore, a

salt-derived anion-rich SEI was found to be formed at the negative electrode, as

illustrated in Figure 1. 8c. Spectroscopy analysis such as X-ray photoelectron

spectroscopy (XPS) and Fourier-transform infrared spectroscopy (FTIR) show that the SEI

is majorly composed of inorganic compounds such as LiF, SO2 or SOx43,45,52–55 or Li2O56.

Some CF3 compounds were also identified as SEI contributor43. However, as pointed out

by Wang et al.57 and Yamada et al.58, imide-based salts tend to decompose upon Ar+

sputtering or X-ray radiation, thus caution must be exercised when performing SEI

surface analysis. Besides, scanning electron microscopy (SEM), transmission electron

microscopy (TEM) and atomic force microscopy (AFM) observations show a dense and

uniform passivation film53–55.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 29

Figure 1. 8 Snapshots of the interfacial structure negatively polarized from molecular dynamic (MD) simulations of (a) a diluted electrolyte based on 1 M LiPF6 in EC:DMC50 (b) a superconcentrated electrolyte based on LiTFSI(DMC)1.2 51. Adapted from Ref50,51. (c) Scheme of the SEI formed in superconcentrated electrolyte. Adapted from Ref54. CIP: contact ion pair. AGG: aggregated cation-anion pairs.

Overall, both changes in bulk properties and interfacial reactivity impact the battery

performances when switching from diluted to superconcentrated electrolytes. Major

contributions have been made in this field, and are summarized in Figure 1. 9. Finally,

one should recall that this topic is not recent. Indeed, more than thirty years ago, the

pioneering work of McKinnon and Dahn59 introduced a propylene carbonate (PC)-based

electrolyte saturated with LiAsF6 salt, enabling to cycle layered electrodes such as LixZrS2

without PC co-intercalation. Subsequent works on superconcentrated electrolytes were

shown to enable cycling graphite electrode while preventing solvent co-

intercalation43,52,53,56,57,60–65. Besides, polymer-in-salt electrolytes, as introduced by

Angell et al.66 in 1993, paved the way for reaching good transport properties in

superconcentrated electrolytes (σ > 0.1 mS/cm at 25 °C in a 9:1 Li salt:polyethylene oxide

(PEO) molar ratio electrolyte). In addition, high charging rates could be reached while

preventing Li dendrites to be formed due to reduction of the cell polarization, thus

greater cycling stability is obtained. Altogether, based on the bulk and interfacial

properties of superconcentrated electrolyte, 5 V-class operation battery were

assembled.

Despite all these advances, the use of organic superconcentrated electrolyte faces

two major issues: the price and the sustainability. Indeed, knowing that in diluted LiPF6-

based electrolyte, the salt represents 10 % of the weight but 70 % of the price of the

electrolyte, one can easily imagine than the cost of superconcentrated electrolytes will

be critical for applications. Additionally, imide-based salts often employed to reach

supersaturation are toxic and corrosive.

30 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 9 Historical timeline of organic-based superconcentrated electrolyte highlighting some of their main properties. All references (Ref43,46,52,53,56,57,59–74) are given in Section References at the end of this manuscript. Adapted from Ref44. Abbreviation of salts and solvents are given at the end of the manuscript.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 31

The development of aqueous rechargeable Li-ion battery (ARLIB) relies on the

compromise between electrode material performances (specific capacity, intercalation

potential and material stability) and the aqueous electrolyte stability. Though ARLIB are

theoretically cheaper, safer, and less toxic than their organic counterparts. Moreover,

they provide high electrolyte conductivity (σ ≈ 100 mS/cm75) and fast reaction kinetics,

thus reducing the overall cell impedance76. Indeed, the activation energy for charge

transfer in aqueous electrolyte was found to be between 23 and 25 kJ/mol, compared

to ≈ 50 kJ/mol in organic electrolyte75. Thus, high rate capabilities77 are achievable, as

well as good cycling stability78 and greater specific capacity79. Besides, despite a

decrease of conductivity in saturated electrolyte due to lower amount of dissociated

charge carriers, the increase in concentration reduces the water activity, as previously

discussed in LiNO3-80–89 and Li2SO4-based electrolyte77,78,90–96, thus kinetically expanding

the ESW89,92 and promoting Li+ intercalation compared to proton intercalation97 (even

though the later has been recently qualified98).

Positive electrodes for ARLIB must operate in a 3 to 4 V vs Li+/Li potential window to

avoid OER, as shown in Figure 1. 10. Therefore, LMO77,78,80,81,90,91,97,99–103 was extensively

used as well as the well-known LCO82–84,92, LFP85,86,93,104,105 or NCM94. Carbon coating was

reported to improve electrode stability91,104 while oxygen removal was shown to

decrease polarization, cell resistance87,106 and self-discharge107. However, proton

intercalation may occur and compete with Li+ intercalation -though depending on the

electrode structure (favorable in layered structure such as LCO and NCM108 compared

to LFP). Proton and lithium co-intercalation is detrimental to cell performances as it

blocks the Li+ diffusion, thus off-balancing the intercalation of Li+ in the electrodes and

decreasing its related capacity. To avoid proton co-intercalation, adjusting pH to values

greater than pH = 9 was shown to improve cell performances109,110.

Though, the negative electrode is even more challenging as the HER potential

(EHER = - 0.06pH V vs ESH, 2.62 V vs Li+/Li at pH = 7) and the absence of SEI formation in

inorganic-based electrolyte58,76,80,111 prevents the use of low potential negative

electrodes such as graphite or metallic Li (see Figure 1. 10). Thus, vanadium-based

(LiV3O8 82,85,99, VO2

80, V2O5 88) or NASICON polyanionic compounds (LiTi2(PO4)3

91,96,112,

LTP) were majorly reported. As a consequence, the specific energy of ARLIB remains

32 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

limited to ≈ 50 Wh/kg with a nominal voltage of 1.5 V, far below that of organic-based

LIB (≈ 200 Wh/kgcell and ≈ 4 V). Also problematic, active materials such as vanadium-

based109 negative electrodes (or LMO95 positive one) are prone to dissolution in aqueous

electrolyte. Moreover, passive components (current collectors for instance) can suffer

from corrosion. Titanium and stainless steel were reported to be the most appropriate

ones109 even though stainless steel still suffer from corrosion in acidic media113.

Therefore, ARLIB as introduced as early as 1994 by Jeff Dahn et al.80 using LiMn2O4 as

positive and VO2 as negative electrode and a saturated LiNO3 electrolyte was never

commercialized.

Figure 1. 10 Pourbaix diagram and intercalation potential of some electrodes used in LIB devices. Adapted from Ref83.

In order to overcome voltage limitation, research efforts were focused on the

development of electrode coating to physically impede water to reach the negative

electrode. To do so, a combination of Li-based gel polymer and ionic conductive ceramic

(LISICON) were used to enable cycling metallic Li in aqueous electrolyte112,114–119, as

illustrated in Figure 1. 11a. Diazonium grafting methods were also developed to

covalently bond polyether-moieties known for their ability to conduct Li+ cations to the

negative electrode material120. The nature of the ether group bounded to the phenyl

ring have a great influence on the electrochemical properties by allowing Li+ transport

while preventing water access to the interface121. However, none of the above-

mentioned strategies were commercially successful.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 33

Figure 1. 11 (a) Scheme of metallic Li coated with a gel polymer electrolyte (GPE) (PVdF: polyvinylidene fluoride, PMMA: poly(methyl methacrylate) and ceramic layer (LISICON). (b) Principle of grafting consisting of the electrochemical reduction of a diazonium salt produced by nitrosylation of an amine by isoamyl nitrite. Adapted from Ref117,120.

Aqueous superconcentrated electrolyte: can the modification of the physico-chemical properties and the interfacial reactivity unlock the competitiveness of aqueous Li-ion batteries?

Developing stable SEIs in aqueous media has been at the forefront of research for

aqueous Li-ion batteries. Indeed, while limitations are found regarding the cathodic

stability (corresponding to the negative electrode side) of classical diluted organic

electrolytes, no stable SEI components are formed in diluted aqueous electrolytes79.

Therefore, based on the promising properties of superconcentrated electrolyte, Suo et

al.111 developed in 2015 an aqueous superconcentrated electrolyte referred to as

Water-in-salt electrolyte (WiSE), quickly followed by the introduction of Water-in-bisalt

(WiBS) in 2016 by Yamada et al.58 and Suo et al.122. The use of superconcentration in

aqueous based electrolyte was shown to enable enlarging the operating potential

window of aqueous systems up to 3 V thanks to the formation of a SEI at the negative

electrode and double layer effect at the positive electrode, while preserving good

physico-chemical properties due to modification of the electrolyte structure, as

discussed in the following section.

34 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Prior to look into greater details to the electrolyte solvation structure, one must

mention that for aqueous superconcentrated electrolyte the salt concentration is

usually express in term of molalities (mol/kg) rather than molarities (mol/L). The Figure

1. 12 shows the corresponding molarities and molalities as function of LiTFSI amount

(the mole and the mass fraction are shown). For instance, 20 mol/kg of LiTFSI

corresponds to ≈ 5 mol/L. Moreover, organic imide-based salts such as LiTFSI are

generally preferred123 as LiTFSI is known to have good electrochemical and chemical

stabilities (towards hydrolysis and in temperature) in aqueous environment124 as well as

keeping good dissociation properties. However, LiFSI is usually avoided as it is prone to

hydrolysis in aqueous environment125.

Figure 1. 12 Molality (mol/kg) (blue crosses) and molarity (orange crosses) as function of mole (bottom x axis) or mass fraction (top x axis). Data extracted from Ref126,127.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 35

In a diluted electrolyte (≈ 1 M), ions are almost fully dissociated and water exists as

“free” water, i.e. as non-ion-bounded molecules forming hydrogen-bounding (H-bond)

network128 (see Figure 1. 13a). When concentration increases, the ion-ion interactions

increase as CIP are formed above 3 M (≈ 10 m), leading to 50 % of the TFSI- anions being

directly in contact with Li+ (see Figure 1. 13b)129. Aggregate structure are observed at

superconcentration above 5 M (≈ 20 m)130, as illustrated in Figure 1. 13 and shown by

Raman spectroscopy in Figure 1. 14a.

Figure 1. 13 Scheme of solvation structure of the electrolyte in (a) diluted electrolyte, (b) concentrated electrolyte and (c) superconcentrated electrolyte. Adapted from Ref131.

As shown in Figure 1. 14a, the intensities attributed to H-bond network and related

to OH stretching at 3255.5 cm-1 and 3403 cm-1 diminish when the concentration

increases from 0 to 5 M (> 20 m). This diminution is concomitant with the appearance

of a sharp peak at 3565 cm-1 suggesting the disruption of the H-bond network and the

absence of free water cluster58,123,132, also determined by molecular dynamic (MD)

calculations128,133. Moreover, 7Li and 19F nuclear magnetic resonance spectroscopy

(NMR) analysis showed changes in ionic structure, as shown in Figure 1. 14b and c. The

lowering of the 7Li chemical shift observed upon increasing salt concentration results

from an increase of the electronic density around Li+, suggesting greater amount of Li+-

anions interactions. A shift of the 19F signal was also observed, suggesting greater anion-

anion interactions134.

36 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 14 (a) Raman spectra in the range 2800-4000 cm-1 observed for LiTFSI aqueous solutions. 2800-4000 cm-1 range corresponds to O-H stretching vibration restricted in a three-dimensional H-bond network structure. Adapted from Ref130. Normalized NMR spectra of aqueous electrolytes showing the chemical shift of water molecules in the presence of LiTFSI at several molalities (1 m blue, 5 m green, 10 m orange and 20 m red). (b) 7Li (c) 19F. Adapted from Ref134.

Moreover, the structure of aqueous superconcentrated electrolytes was reported to

be an anion TFSI--rich domain that behaves as an immobile matrix in which a water-rich

domain (network domain135 or cluster one136) is interpenetrated, as depicted in Figure

1. 15. Fast Li+ transport is enabled through the water-rich domain thanks to bulk-like

water structure, i.e. water which forms channels to support fast transport, intertwined

in the porous TFSI--rich skeleton137 and interfacial water, i.e., water bounded to Li+ cation

that move through the water bulk-like channels. Though, Li+ transport may also take

place through the TFSI--rich domain136. Altogether, these two networks give a

heterogeneous structure with asymmetric clusters. Based on the identification of these

two domains, the idea that upon increase in salt concentration water is not displaced

from the first solvation sheath is strengthened, suggesting that all water molecules will

preferentially bounds Li+ cation to form Li(H2O)4+ clusters rather than be involved in a

mix first solvation sheath composed of TFSI- anion and water.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 37

Figure 1. 15 Schematic diagram of Li+ cation transport in aqueous superconcentrated electrolyte composed of two interpenetrating network: a water-rich one (orange) and an anion-rich one (green/dark-blue). Li+ cationic transport mechanism remains unclear between vehicular mechanism and hoping-type one. Adapted from Ref138.

First, as ionic association is enhanced in superconcentrated electrolytes, the number

of dissociated charge carriers decreases, thus the conductivity, as shown in Figure 1.

16a. However, conductivity values remain rather high (≈ 10 mS/cm in 21 m LiTFSI111)

thanks to the heterogeneous structure previously described that enables fast Li+

transport and provides high Li+ transference number139 (> 0.6135,140,141 and found as high

as 0.73135 depending on the experimental or theoretical method of determination). The

difference in transference number between Li+ and TFSI-, considered as immobile, is

related to the asymmetry between anion and water clusters140. Moreover, as viscosity

is impacted by the concentration and the interaction between ions and solvent

molecules142, its value is found to increase to values greater than 40 mPa.s in WiSE (see

Figure 1. 16.b). Besides, as illustrated in Figure 1. 16c on the Walden plot,

superconcentrated aqueous electrolytes reach the “ideal KCl line” and may even

become superionic solution, i.e., when conductivity and viscosity are decoupled,

especially considering bi-salt media58 which reach concentration above saturation, thus

explaining that good transport properties are preserved.

38 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 16 (a) Conductivity measurements for LiTFSI aqueous solutions as function of concentration. (b) Relative viscosities measured for LiTFSI aqueous solutions as function of concentration. Adapted from Ref139. (c) Walden plot for LiTFSI solutions as function of concentration. The Walden plot was generated from the molar conductivities (Λ) and the viscosity (η) of the electrolyte solutions. In a Walden plot, electrolyte solutions can be classified in terms of their performances as ionic conductors: superionic (upper left region above the ideal KCl line) which states a decoupling between viscosity and conductivity, good-ionic (on the ideal line), poor-ionic (bottom right region under the ideal line), or non-ionic (far below the ideal line) liquids. For the LiTFSI/H2O solutions, the plot approaches the ideal line with increasing concentration, and finally joins with the ideal line at saturation. “Superionic” solution are reported for WiBS. Adapted from Ref58,130.

Moreover, the increase of ESW, especially on the positive electrode side, was

partially attributed to a diminution of water reactivity58,134 related to a decrease in water

activity58. Indeed, alike in organic superconcentrated electrolytes, as water is majorly

bounded to ions, the energy required to desolvate water and thus to oxidize it is greater,

thus pushing the onset potential of OER to higher potential143. As an example, the vapor

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 39

pressure of water was measured to decrease from 4.25 kPa for pure water to 0.50 kPa

at 30°C in Li(TFSI)0.7(BETI)0.3 electrolyte.

However, considering the negative electrode side and thus the HER parasitic reaction,

water molecules bounded to Li+ are more reactive than “free” water molecules upon

reduction, as found by Dubouis et al.144. Moreover, greater activity for protons was

recently reported in superconcentrated electrolyte98. These results suggest that the

overall decrease of water activity is not sufficient to effectively increase the ESW,

especially on the negative electrode side. Finally, upon increase of concentration,

electrolytes become more acidic, thus shifting the onset potential of HER and OER to

greater potential. Eventually, this increase was found to be similar to the increase in Li+

insertion potential resulting from the increase in Li+ activity as function of salt

concentration58. Therefore, shifting to WiSE does not favor Li+ intercalation at the

negative electrode relative to HER, for instance. One has to note that caution must be

taken while conducting pH measurements in superconcentrated electrolyte as pH-meter

with glass probe should be avoided125. Acidification of the electrolyte was found with

pH value as low as 2.8145, even though a recent study determined that such drop was

overestimated and that less than 1 mM of protons are formed in WiSE98.

Eventually, WiSE cannot usually be used at low temperature due to their rapid

crystallization138. Indeed, as shown on the phase diagram (see Figure 1. 17, purple line

for WiSE-region), the liquidus temperature of WiSEs is generally around 20 °C.

40 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 17 Liquid-solid phase diagram of LiTFSI-H2O binary system. Adapted from Ref126. The purple line represents the 18.5-21 m LiTFSI concentration considered for aqueous superconcentrated electrolyte.

At the negative electrode side, the origin for the enlarged electrochemical window

(see Figure 1. 18) offered by aqueous superconcentrated electrolytes is ascribed to the

anion degradation that forms a LiF-rich passivating SEI that pushes the HER onset

potential from 2.62 V vs Li+/Li (at pH = 7) in 1 m LiTFSI to 1.9 V vs Li+/Li111 in 21 m LiTFSI

(see Figure 1. 18a). The exact mechanism for the SEI formation is still under debate,

with three mechanisms being proposed, and further detailed in the appendix of the

chapter: (i) the direct electrochemical reduction of anions or anion clusters such as

Li2(TFSI)+ 58,111,146–148, (ii) the chemical degradation of anion by nucleophilic attack

resulting from the HER reaction and the generation of hydroxyls134,149, (iii) the

precipitation/dissolution of LiTFSI salt149. Besides, dissolved gases (O2 or CO2) were also

reported to be involved in SEI formation. Though, despite the differences in SEI

formation mechanism, all are combined with the lower solubility of SEI compounds in

WiSE148.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 41

Figure 1. 18 Electrochemical stability window reported by Suo et al.111 in 2015 for several electrolyte ranging from 1 m to 21 m LiTFSI on stainless steel. (a) Zoom on the cathodic stability at the negative electrode side. (b) Zoom on the anodic stability at the positive electrode side. (c) Overall ESW.

Nonetheless, water enrichment following the formation of hydrated cations

([Li(H2O)x]+) is still found below 1.5 V vs Li+/Li Ref 150,151 at the negatively charged interface,

with hydrogen atoms oriented perpendicular to the surface, thus promoting the HER150.

Figure 1. 19 illustrates this enrichment near the negative electrode, often referred to as

the cathodic challenge.

42 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 19 (a) Contributions of ions and water to the atom density of the interfacial layer as a function of applied electrode potential assessed by classical molecular dynamics (MD). Blue region represents the interface of a negatively charged electrode mostly populated by water molecules. (b) Cumulative atom number density profiles normalized by bulk density as a function of distance from the electrode (z) and snapshots of the interfacial layer at -2 V vs the PZC (b). Adapted from Ref150.

Moreover, the increase in ESW stability is in reverse order with the intrinsic HER

electrocatalytic activity of the electrode material used to measure it152,153, as illustrated

in Figure 1. 20.

Figure 1. 20 Linear sweep voltammograms for 21 m LiTFSI on platinum, gold, stainless steel (SS), titanium, glassy carbon (GC), and aluminum. The pH of all solutions was adjusted to a value of ≈ 5. Experiments were carried out at room temperature. Adapted from Ref153.

As a conclusion, the SEI stability can be found highly dependent on the electrode

material, the anion chosen and the applied potential148,152–154. Last but not least,

regarding the assessment of the ESW and thus the interfacial reactivity, the ESW can be

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 43

artificially increased by choosing a high threshold of current density (ithreshold). For

example, the ESW increases from 2.1 V to 3.1 V by selecting a ithreshold of 100 µA/cm²

rather than 2 µA/cm² 153.

Turning to the positive electrode, a double layer effect enables preventing the OER.

Indeed, upon positive polarization and in superconcentrated electrolyte, the negatively

charged anion species move towards the positive electrode to form an hydrophobic

organic anion-rich double layer repealing water molecules, thus preventing their

oxidation (through OER)150,152,155 (see Figure 1. 21). Hence, a shift in OER onset potential

from 3.85 V vs Li+/Li in 1 m LiTFSI at pH = 7 to 4.9 V vs Li+/Li in 21 m LiTFSI was

observed111. The decrease of fraction of free water molecules was also proposed to

diminish the amount of water that can be oxidized150, though such consideration does

not consider the strong impact of the electrocatalytic activity of the electrode material.

Moreover, the anion choice impacts the interfacial reactivity at the positive electrode

side. For instance, TFSI- anion is preferentially adsorbed over Otf- anion at the

electrode/electrolyte interface150 leading to a more effective screening even though Otf-

anions, that decompose at lower positive potentials, may form a partially insulating

CEI156.

Figure 1. 21 (a) Contributions of ions and water to the atom density of the interfacial layer as a function of applied electrode potential assessed by classical molecular dynamics (MD). Pink region represents the interface of a positively charged electrode mostly populated by anions. (b) Cumulative atom number density profiles normalized by bulk density as a function of distance from the electrode (z) and snapshots of the interfacial layer at +2 V vs the PZC. Adapted from Ref150.

44 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

As described above, the electrolyte structure of WiSE impacts both bulk and

interfacial properties, and thus the battery performances. Hence, the use of aqueous

superconcentrated electrolyte was proposed to enable cycling negative electrode

materials which Li+ insertion potential lays outside the ESW of water or diluted aqueous

electrolyte (LTO, Mo6S8, TiO2, TiS2, see Figure 1. 10). In Figure 1. 22 is shown the

discharge capacity and the Coulombic efficiency of WiSE-based batteries. As a proof of

concept, a 2.3 V battery using Mo6S8 and LiMn2O4 as negative and positive electrodes,

respectively, was first reported in 2015 by Suo et al.111. Following this demonstration,

Yamada et al.58 then showed that mixing two organic lithium salts

(Li(TFSI)0.7(BETI)0.32∙H2O) increases further the concentration and enables assembling

aqueous batteries with a working potential as high as 3.1 V using LTO in combination

with LiNi0.5Mn1.5O4 or LCO electrodes. Moreover, WiBS electrolyte (21 m LiTFSI : 7 m

LiOtf) was also employed by Suo et al.122 to assemble a 2.5 V TiO2/LiMn2O4 battery.

Furthermore, using TiS2 as negative, a 1.7 V TiS2/LiMn2O4 cell using a 21 m LiTFSI

electrolyte was reported157. However, it was also observed that using LTO as negative

electrode leads to drastic decrease in cell performances (see Figure 1. 22d, from cycle

number 50).

Figure 1. 22 Cycling stability of several aqueous superconcentrated electrolyte based on LiTFSI-salt. (a) Mo6S8/LiMn2O4 in 21 m LiTFSI at 0.15C. (b) TiS2/LiMn2O4 in 21 m LiTFSI at 1C. (c) TiO2/LiMn2O4 in 21 m LiTFSI and 21 m LiTFSI : 8 m LiOtf at 0.5C. (d) Li4Ti5O12/LiNi0.5Mn1.5O4 in 20 m LiTFSI : 8 m LiBETI at 0.5C. The cell capacity is calculated based on the total weight of the positive and negative active materials. Adapted from Ref58,111,122,157.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 45

Apart from improving capacity retention and Coulombic efficiency, the use of WiSE is

said to impact solubility equilibrium of SEI or transition metal due to a decrease in water

activity related to lower polar properties of water as most of water molecules are

involved in the solvating sphere158. Moreover, both SEI and electrode dissolution rates

were reported to kinetically decrease as the increase in viscosity reduces the diffusion

of dissolved species in the bulk and confine the dissolved species near the interface. At

the positive electrode side, the smaller electrode area exposed to water due to the

population of the double layer by anions158 also kinetically reduce electrode dissolution

rate. When formed by using Otf- anions or additives such as tris(trimethylsilyl) borate

(TMSB), CEI can also prevent transition metal dissolution such as Co or Mn at the

positive156,159.

Moreover, as pH of WiSE is neutral or mildly acidic, the use of Al current collector at

the negative is enabled. Indeed, Al passivation domain is comprised between pH values

of 4 to 8.5. At the positive electrode, the repulsion of water from the double layer,

combined with the high TFSI- oxidation potential, slows down the kinetics of Al

dissolution160, therefore enabling its use as current collector150. Altogether, the

possibility to use Al current collectors on both sides is of great interest as it has a low

density (light weight), high electronic conductivity, low cost and great ability to process

thin rolls.

As a conclusion, these studies have renewed interest for aqueous systems relying on

the use of superconcentrated electrolytes, which was later on extended to aqueous Na-

ion161–165, K-ion166–169, Zn-ion170,171 based on the promises of extending the ESW, Li-O2172

in which reversible Li2O2 formation is observed unlike in diluted aqueous electrolytes,

Li-S173,174 in which WiSE prevents from polysulfide redox shuttle, Zn metal-based cell

showing reversible Zn plating stripping170,175,176 or dual-ion battery including the halogen

conversion-intercalation177–179. However, several observations such as the fast capacity

decay when using LTO as negative electrode call for further investigations to understand

if the improved of ESW in WiSE is real in practice and to improve the overall

performances of these systems, especially regarding instabilities at the negative

electrode (i.e. the cathodic challenge).

46 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Further improving aqueous superconcentrated-based LIB by expanding the ESW limit at the negative electrode side

Several strategies are employed to solve the cathodic challenge. On the one hand,

the tuning of the electrolyte properties with the removal of free water molecules from

the electrolyte149, which would potentially open the path for increased ESW, or the use

of additive to suppress water from Li+ solvation sheath, as reports pointed towards the

greater reactivity of water molecules participating to the Li+ solvation sheath compared

to free water molecules144 are envisioned. On the other hand, several research groups

focused their efforts on the use of coatings capable of preventing water to access the

interface.

In binary mixtures, the electrolyte concentration is found to increase to greater

values than the solubility limit of each of the two salts, thus reducing the amount of free

water. Indeed, a hydrated salt can dissolve a non-hydrated salt which possess similar

chemical properties. Mix-anions WiBS electrolytes58,122,180,181, i.e. two salts based on the

same cation but different anions, were first introduced in 2016 by Suo et al.122 and

Yamada et al.58, using 21 m LiTFSI : 7 m LiOtf and Li(TFSI)0.7(BETI)0.3 2∙H2O, respectively.

These systems are generally based on the use of stable chaotropic anions, i.e. disrupting

the bulk-like water interactions, and good SEI-former anions, as rationalized by Reber et

al.182. However, despite the decrease of free water molecule in the bulk, the increase in

ESW for mix-anion systems is very limited, if not inexistent at the negative electrode

(see Figure 1. 23).

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 47

Figure 1. 23 Electrochemical stability window of 20 m LiTFSI and 20 m LiTFSI : 8 m LiBETI on stainless steel. Linear sweep voltamogram performed at 0.1 mV/s at room temperature. Adapted from Ref153.

Mix-cation electrolytes183,184, i.e. two salts based on one anion but using two different

cations (Na-Li, K-Li or Na-K-based ones), enable to achieve greater solubility than mix-

anion162, further minimizing the water to salt ratio. Nonetheless, a major drawback of

this strategy is the co-intercalation184 of both alkali cations that leads to fast

performances decay. Thus, ammonium inert co-cations (such as tetraethylammonium

(TEA+) or trimethylethylammonium (Me3EtN+)) with larger radii were used162,183 (see

Table 1. 1, for a comparison of cation radii).

Table 1. 1 Radii of different cations used in WiBS.

Besides, ternary electrolytes -based on the introduction of ionic liquid185 (IL) for

instance- have also been used to further increase the solubility of salts, as the entropy

of mixing increases more in ternary than in binary mixtures. Eventually, to increase the

Alkali metal Li+ Na+ K+ TEA+

Ionic radius (Å) 0.68 0.97 1.33 3.37

48 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

solubility limit and the thermal stability by shifting the liquidus temperature to lower

temperature163,165,186, the high vibrational mobility and flexibility of asymmetric

moieties of ions is useful as they impede the formation of long range order165,180,181,187.

The asymmetry of the anion, such as FTSI- or PTFSI-, enable the formation of an

asymmetrical solvation sheath of Li+ (see Figure 1. 24a, red circle). Hence,

uncoordinated moieties, freer and more mobile, rotate and disturb the surrounding

solvation structure preventing close packing of anions and cations (see Figure 1. 24a,

purple circle). Furthermore, the asymmetry of the anion itself reduces the probability

for specific rearrangement.

Figure 1. 24 (a) Schematic illustration of the difference in local coordination for symmetric TFSI- and asymmetric FTFSI- anion. (b) The numbers of hydrogen bond and coordination to cations around a water molecule, which were obtained by averaging three trajectories (The geometric criteria of hydrogen bonds were defined by the radial distance between the donor and acceptor oxygen atoms (< 3.5 A) and the angle between the acceptor oxygen atoms and a donor-H-acceptor (135–180°) for a diluted 1 M LiTFSI and a WiBS electrolyte based on Li(PTFSI)0.6(TFSI)0.4). Adapted from Ref181,186.

However, even though by using mix-anion, mix-cation or asymmetric ion-based

electrolytes all water molecules are generally bounded to Li+ with negligible H-bonds, as

calculated by MD simulation for mix-asymmetric anion electrolyte (see Figure 1. 24b),

they generally leads to a decrease in conductivity and an increase in viscosity, as shown

in Figure 1. 25 (bottom part) with limited thermal operating range (top part) and

restricted increase in the cathodic limit of the ESW (middle part), restraining the use of

negative electrode to LTO at best. Therefore, while research efforts dedicated to novel

salts need to be continued to further improve electrolyte properties, other solutions are

also investigated. Among them, tuning the solvation sheath of the Li+ cations by using

hybrid organic/aqueous electrolyte to remove water from the first solvation sheath of

Li+ and thus prevent water reduction was proposed.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 49

Figure 1. 25 Summary of some of the electrolyte properties (liquidus temperature, conductivity, viscosity, water to salt molar ratio, cathodic limit of the ESW (corresponding to the negative electrode side)) using mix-anion (dark blue), mix-cation (light blue), asymmetric-imide anion (pink) and asymmetric cation (underlined) in Li-ion battery. Adapted from Ref58,111,122,180,181,183–185,187.

50 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

First, hybrid aqueous/non-aqueous electrolyte reduces water activity in the bulk by

forming H-bonds between non-aqueous solvent and water, as observed in

dimethylsulfoxide (DMSO)-based electrolyte188,189. Moreover, the addition of co-solvent

strongly coordinating Li+ can promote the removal of water from the first solvation

sheath of the alkali ion, or at least decrease its quantity (see Figure 1. 26a), thus

preventing or slowing down water reduction. DMSO, dimethylformamide (DMF) and

urea188–190 were identified as good candidates owing to their greater donor numbers

than water (29.8 for DMSO vs 18 for water). On the contrary, DMC and acetonitrile

addition191,192 impacts the solvation sheath by increasing the fraction of water in Li+ first

solvation sheath, thus promoting the nanophase separation leading to a water-rich

domain and a TFSI-DMC-rich domain, as shown in Figure 1. 26b and c. This promoting

effect will be detrimental for the cathodic stability of WiSE.

Figure 1. 26 (a) Modification of the solvation sheath of Li+ cation upon addition of urea. Adapted from Ref190. (b) Snapshots (including front and right side views) of inner-Helmholtz layer of negatively charged electrodes obtained from molecular dynamics (MD) simulation in 15.3 m LITFSI in water:acetonitrile (1:1 molar ratio). (d) Proportion of the Li+ solvating with 1, 2, 3 and 4 solvent molecules (H2O or DMC), obtained from molecular dynamic (MD) simulations. Adapted from Ref192.

Furthermore, not only the reduction potential of the co-solvent needs to be greater

than the HER to promote the SEI formation, but the reduction product also has to be

insoluble. Besides, as neutral solvents are less sensitive to negative repulsion, their

reduction should be easier than TFSI-based clusters, thus leading to their participation

to the SEI formation. DMC192, acetonitrile192, tetraethylene glycol dimethyl ether193

(TEGDME), polyacrylamide194 (PAM) reduction during first charges were reported to

contribute to the presence of organic components in the SEI, thus enabling to expand

the cathodic limit of ESW below 1 V vs Li+/Li at the negative electrode while LiF is kept

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 51

in the inner layer due to TFSI- degradation192 (see Figure 1. 27a and b). Upon urea, 1,5-

pentanediol195 or sulfolane196 addition, an organic-inorganic SEI was also identified as

illustrated by the dense SEI majorly composed of Li2CO3 and amorphous polyuria in urea-

based electrolyte (see Figure 1. 27c) but without LiF participation and limited increase

in the cathodic limit of the ESW (see Figure 1. 28).

Figure 1. 27 (a) TEM images of LTO negative electrode after cycling showing the formation of a LiF-based SEI in 15.3 m LiTFSI in H2O:acetonitrile. (b) Intensity changes of XPS analysis of LiF (yellow), C≡N (dark red), R-S-N-S (red) from F 1s and N 1s with various sputtering durations (c) TEM images of Mo6S8 negative electrode after 10 cycles showing the formation of a Li2CO3 SEI in 1:3:2 LiClO4:H2O:urea electrolyte.

Bulk properties, notably the conductivity, the cathodic limit of the ESW

(corresponding to the negative electrode side) and the thermal properties are reported

in Figure 1. 28. The expansion of the ESW enables to cycle LTO negative electrode in a

mixture of water:acetonitrile at 0.2C193, using LiTFSI:H2O (1:2.8 molar ratio) and

LiTFSI:TEGDME (1:0.41 molar ratio) in the proportion 1:1 (mass ratio), though only

charging rates faster than 1C were tested193 or in the sulfolane:water-based

electrolyte196. Other electrolytes are reported to cycle with negative electrode having

greater intercalation potential, such as Mo6S8190,195 or TiO2

194. Moreover, as observed in

Figure 1. 28, the expansion of the ESW is in reverse order with the enhancement of

conductivity (often lower than ≈ 3 mS/cm). Thus, tuning the solvation sheath by adding

non-aqueous solvent, which can be related to the concept of diluted-concentrated

electrolyte, may improve the cell performances by enabling the cycling of LTO electrode

but this approach does not entirely overcome the cathodic challenge yet neither

preserve good physico-chemical properties.

52 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Figure 1. 28 Summary of some of the electrolyte properties (conductivity, viscosity, cathodic limit of the ESW and thermal stability) using solvation sheath tuning in Li-ion battery. Adapted from Ref190–196.

The incorporation of WiBS in a polymer matrix enables to reduce water activity197 by

coordinating water with the polymer moieties. UV-curable gel-polymer electrolyte (GPE)

(see Table A.1. 3, at the end of the chapter for details) using low viscosities polymers

was proposed to enable good wetting of the porous electrodes197–199. Moreover, in the

spirit of suppressing the use of fluorinated salts and developing low-cost electrolyte, He

et al.200 reported a Water-in-ionomer electrolyte based on 50 wt% LiPAA but

constraining the negative material selection to the use of TiO2. Furthermore, the

suppression of the classical H2O-H2O H-bonds structure and the reduction of the amount

of free water is promoted upon addition of PEG201,202 or PEO203. Such strategy enables

decreasing the concentration of LiTFSI down to 2 m LiTFSI in a 2 m LiTFSI∙PEGx(H20)(1-x),

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 53

71 < x (wt%) < 94, leading to a decrease in cost and toxicity, as reported by Xie et al.201

and Li et al.202.

However, alike for classical WiSE, a Li(H2O)4+-rich domain and a polymer-LiTFSI-rich

domain are found, suggesting that the cathodic challenge remains unsolved203. Besides,

in these GPE, alike previously observed for organic/aqueous electrolytes, the ESW and

the transport properties were found to be directly related to the water content201,203.

Indeed, increasing the water concentration from 6 wt% to 29 wt% in a LiTFSI-PEG-based

GPE enhances the conductivity from 0.9 mS/cm to 3.4 mS/cm, while unfortunately

reducing the ESW by ≈ 500 mV201.

Eventually, in 2017, Yang et al.204 reported the assembly of batteries using metallic Li

or graphite in combination with high potential positive electrode using a WiBS-gel

polymer based on 21 m LiTFSI : 7 m LiOtf with 10 wt% of polyvinyl alcohol (PVA).

Nonetheless, these outstanding performances rely more on an extra organic polymer

coating layer used to protect the negative electrode from the HER, as shown in Figure 1.

29, than on the use of WiBS.

Figure 1. 29 Electrochemical stability window (ESW) of WiBS-gel polymer with or without an extra organic coating. Adapted from Ref199.

The use of metallic Li or graphite in WiSE-based electrolyte were shown to be enabled

by the use of organic-coatings that (i) provides good mechanical properties, (ii) prevents

54 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

water from accessing the interface and (iii) may contribute to the SEI formation. The first

type of organic coating integrates a GPE consisted of a HFE-LiTFSI-PEO GPE (HFE stands

for highly fluorinated ether, here 1,1,2,2-Tetrafluoroethyl 2,2,2-Trifluoroethyl Ether)204.

During first charges, the GPE undergoes reductive decomposition to form a LiF-based

SEI which properties are enhanced by the contribution from organic-based compounds.

Using this system, graphite and metallic Li electrodes were cycled 50 cycles at 0.3C with

≈ 99 % of Coulombic efficiency. Similarly, a UV-induced GPE coating using 1 m LiTFSI in

fluoroethylene carbonate:trifluoroethyl methyl carbonate (FEC:FEMC, 1:1 vol%) was

found to passivate the graphite electrode199. However, the graphite/LCO cell

performances show a rapid capacity decay (see Figure 1. 30). Besides, as proposed by

Dubouis et al.134, TFSI- anions undergo a chemical degradation in presence of HO- anion

to form the SEI. Therefore, Zhang et al.198 cycled a LTO-based cell for 200 cycles at 0.5C

using a strongly basic solid polymer electrolyte (SPE), LiTFSI-PEO-KOH, to enable the

formation of a LiF-Li2CO3-containing SEI that incorporates polymeric decomposition

products.

Figure 1. 30 (a) Scheme of a cell based on an organic coating. (b) Capacity and Coulombic efficiency as a function of cycle number for a graphite/LCO cell cycled in hybrid organic/aqueous electrolyte and using a polymer coating as a protection of the negative electrode. Adapted from Ref199.

Inorganic coatings were also tested in WiSE. The propensity of a coating to suppress

or, at least, reduce HER depends on its intrinsic electrocatalytic activity205. Aluminum

oxide (Al2O3) coating was one of the most widely used205–207 as it shows the lower HER

activity (see Figure 1. 31a.) and it is also known to be insoluble in water. However, LTO-

Al2O3-coated electrodes were found to initially deliver 84.5 % of Coulombic efficiency in

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 55

a LTO-Al2O3/LMO battery, suggesting that even though a conformal inorganic coating is

deposited, some defects are present causing some electrolyte consumption. After 60

cycles, the LTO-Al2O3/LMO cell could deliver 99 % of Coulombic efficiency even though

a smooth capacity decay is observed (see Figure 1. 31b). The use of Al2O3 coating was

also reported to suppress the oxygen reaction reduction (ORR)207 and thus enable the

use of open-air cells (see Equation 1. 6 for ORR reactions) by suppressing self-discharge

caused by the presence of dissolved O2107 (Figure 1. 31c). However, such design restricts

the use of WiSE to LTP-like negatives.

Figure 1. 31 (a) Cathodic limits evaluated by linear sweep voltammetry on LTO surfaces coated with different materials (inset shows the enlarged view). Counter electrode: activated carbon; reference electrode: Ag/AgCl, scan rate: 1 mV/s. (b) The cycling performance of the full cell using an Al2O3-coated LTO negative and LMO positive in 21 m LiTFSI. Adapted from Ref205. (c) Effects of the ORR on the self-discharge of the lithiated Li3Ti2(PO4)3 electrodes in a three-electrodes open-cell configuration. The open-circuit potential curve of Li3Ti2(PO4)3 in the 1m Li2SO4 over 10 h of relaxation at open-circuit (orange). The open-circuit potential profile of the lithiated Al2O3@Li3Ti2(PO4)3 electrode in 21 m LiTFSI : 8 m LiOtf (yellow) in the 28m WiSE over 10 h of relaxation at open-circuit. These tests were conducted in an open-cell configuration with exposure to the ambient air. Adapted from Ref207.

56 Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB

Equation 1. 6:

𝑂2 + 𝐻2𝑂 + 2 ∙ 𝑒− = 𝐻𝑂2− + 𝑂𝐻−

𝐻𝑂2− + 𝐻2𝑂 = 𝐻2𝑂2 + 𝑂𝐻

− 𝐻2𝑂2 + 2 ∙ 𝑂𝐻− + 2 ∙ 𝐿𝑖+ = 𝐿𝑖2𝑂2 + 2 ∙ 𝐻2𝑂

Moreover, AlF3-Al2O3206, LTP208

and carbon 122,208,209 coatings were also reported but

limiting the negative electrode choice to the use of TiO2 electrode, far above graphite or

even LTO ones.

Altogether, polymeric coating was reported to be the most efficient strategy as only

this strategy enabled to cycle metallic Li or graphite negative electrodes.

Conclusion of the chapter

The introduction of superconcentrated aqueous electrolytes was proposed to

overcome safety and environmental issues while keeping high performances. First,

superconcentration enables to increase the ESW. Indeed, thanks to the formation of a

SEI, majorly based on salt-derived compound LiF, negative electrodes with insertion

potential laying outside the usual ESW (< 1.5 V) were implemented. Besides, WiSE-based

battery enables to limit electrode and SEI dissolution while maintaining fast Li+ transport

owing to a high Li+ transference number. As summarized in Figure 1. 32, many

researches and advances have been done since the introduction of the seminal 21 m

LiTFSI WiSE electrolyte developed by Suo et al.111 in 2015, though achieving

performances competitive with commercial-LIB still remains to be shown. Starting from

binary or ternary liquid mixtures using mix-anion or mix-cation with asymmetrical ions,

moving to hybrid aqueous/non-aqueous electrolyte to tune the first solvation sheath of

Li+, using gel-polymer type electrolyte to reduce water activity or using artificial

inorganic SEI on the negative electrode to prevent water to reach the electrode, all were

proposed to partially overcome some limitations.

Chapter 1 – Introduction to aqueous superconcentrated electrolyte and their use in LIB 57

Figure 1. 32 Summary of the innovations developed to improve seminal-21 m LiTFSI aqueous superconcentrated electrolyte.

However, improvements previously observed cannot be directly transposed to

practical devices as the increase of the ESW is strongly related (i) to the electrode

material on which it is assessed and (ii) to the current threshold chosen. Eventually, one

must look not only at the ESW but also at the physico-chemical properties to ensure high

conductivity, low viscosity and wide thermal stability. As a conclusion, many parameters

must be taken into account to assess the performances of one electrolyte. Therefore,

the aim of this thesis is to explore the viability of 20 m LiTFSI as an example of aqueous

superconcentrated electrolyte behavior by assessing key parameters representative of

this chemistry to compete with classical LIB. Thus, Chapter 2 focuses on the

performances obtained during cycling and self-discharge while monitoring parasitic

reactions. Chapter 3 will investigate the instability of the native SEI and its ability to

prevent the HER. Eventually, Chapter 4 will assess the use of LiF-based coating as

artificial SEI for WiSE-based battery.

CHAPTER 2 – CYCLING VIABILITY OF AQUEOUS SUPERCONCENTRATED

ELECTROLYTES BASED ON 20 MOL/KG LITFSI AND 20 MOL/KG LITFSI : 8 MOL/KG

LIBETI 1

1 This chapter is based on the article that I co-authored: Droguet, L.; Grimaud, A.; Fontaine, O.; Tarascon, J. Water‐in‐Salt Electrolyte (WiSE) for Aqueous Batteries: A Long Way to Practicality. Adv. Energy Mater. 2020, 10 (43), 2002440. https://doi.org/10.1002/aenm.202002440.

60 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Introduction

As described in Chapter 1, the use of superconcentrated aqueous electrolytes was

proposed to widen the electrochemical stability window (ESW) of aqueous-based

electrolytes. Among them, 21 mol/kg (m) LiTFSI Water-in-salt (WiSE) has been widely

used111 since the seminal publication by Suo et al.111 in 2015 reporting an increase in

ESW up to 3 V. Besides, greater salts concentration -thus lesser water content- were

reached by using Water-in-bisalt (WiBS) electrolyte such as Li(TFSI)0.7(BETI)0.3,

potentially enlarging the ESW even more58. Throughout this chapter, two similar

electrolytes (20 m LiTFSI and 20 m LiTFSI : 8 m LiBETI) will be benchmarked as

representative to study aqueous superconcentrated electrolytes practical viability in Li-

ion battery (LIB).

Prior to study the effect of cycling parameters on performances of WiSEs-based

aqueous batteries, proper current collectors must be selected. The ESW widening was

previously assessed using metallic current collectors such as platinum, conductive glassy

carbon or directly with current collectors materials (titanium, stainless steel or

aluminum) with overpotential for the oxygen evolution reaction (OER) greater than 500

mV measured on the oxidation part58,152,160,161,184,210 using superconcentrated

electrolytes. Instead, almost no change is observed for hydrogen evolution reaction

(HER) overpotential (on the reduction part) as function of the metallic current

collector152,210 when increasing the salt concentration, the exception being aluminum

that passivates58,148,153,154,160. However, these potential shifts are determined by cyclic

voltammetry measurements rather than by potentio/galvano-static methods, hence

departing from practical conditions. Indeed, by narrowing down the number of testing

parameters, especially when the threshold current density is not taking into account,

the influence of parasitic reactions such as the HER can be downplayed, as discussed by

Kühnel et al.153 Figure 2. 1 shows the ESW of 20 m LiTFSI on glassy carbon, aluminum,

stainless steel and titanium current collectors. Aluminum and glassy carbon show an

ESW reduction limit at 1.8 V vs Li+/Li for HER (considering a ithreshold = 0.25 mA/cm²)

followed by stainless steel at 2 V vs Li+/li and titanium reduction limit at 3 V vs Li+/Li

(Figure 2. 1b). On the oxidation limit, titanium passivation enables the increase of the

ESW to potentials greater than 4.5 V vs Li+/Li. Therefore, based on both these results

and their relative ease-of-use, stainless steel current collectors were selected for

electrodes which lithium insertion/de-insertion potential lays in the range 2 to 4 V vs

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 61

Li+/Li, and titanium as current collector electrode materials which Li insertion/de-

insertion potential is above 4 V vs Li+/Li.

Figure 2. 1 Electrochemical stability window of 20 m LiTFSI electrolyte assessed at 100 mV/s on glassy carbon (blue), aluminum (orange), stainless steel (yellow) and titanium (green) as working electrodes (WE), Pt wire as counter electrodes (CE) and Ag/AgCl as reference. In (a) is display the full ESW and in (b) and (c) a zoom on the cathodic and anodic stability part, respectively, are provided.

To independently assess the parasitic reactions occurring at the negative electrode

from those at the positive electrode, Mo6S8 negative electrode was chosen as cathode

while NMC622 was chosen as anode to test the ESW. Mo6S8 and NMC622 electrochemical

signatures are shown Figure 2. 2a. One can notice the classical electrochemical signature

of Mo6S8 with four Li+ ions being reversibly inserted, three at 2.3 V vs Li+/Li and one at

2.7 V vs Li+/Li. Moreover, LFP and LTP were selected as counter electrode for Mo6S8 and

NMC622, respectively. Indeed, using a 3-electrodes Swagelok cell with an Ag/AgCl

reference electrode, both LFP and LTP counter electrodes, known to reversibly exchange

Li+ in aqueous electrolytes76,107,112, were found to have their redox potential within the

ESW of the WiSE electrolyte studied in this work (see Figure 2. 2b). Having defined the

proper current collectors and active materials, full cells were thus assembled to study

62 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

the effect of WiSE on the redox properties of both active materials, Mo6S8 and NMC622,

measured against LFP and LTP, respectively (see Figure 2. 2a). Besides, to avoid that LFP

and LTP counter electrodes limit the cell cycling, capacity ratio were set to 4:1 for

LFP/Mo6S8 and 1.1:1 for LTP/NMC622. Eventually, after checking the redox potentials for

LFP and LTP versus the potential of Ag/AgCl reference electrode, the potentials for the

working electrodes were rescaled versus Li+/Li. Doing so, an upshift in ≈ 230 mV is

observed when cycling Mo6S8 and NMC622 in WiSE compared to the organic electrolyte

(1 M LiPF6 in EC:DMC, e.g. LP30). This shift, previously observed, was assigned to the

effect of the salt concentration on the redox potential of the intercalation electrodes58.

Such an upshift of the intercalation potential combined with the use of a fixed cutoff

potential of 4.2 V defined vs Li+/Li in charge explains the lower measured capacity for

NMC622 in WiSE as opposed to non-aqueous electrolytes (see Figure 2. 2a, orange full

and dash line). The importance of adequately selecting this cutoff potential will be

discussed in greater detail in Section 4.

Figure 2. 2 (a) Galvanostatic charge and discharge signatures for electrode materials. Galvanostatic experiment performed at 1C with Mo6S8 measured in LP30 versus metallic Li and 20 m LiTFSI versus LFP on SS current collector (1st cycle). Galvanostatic experiment performed at 0.10C with LiNi0.6Mn0.2Co0.2O2 measured in LP30 versus metallic Li and 20 m LiTFSI versus LTP on Ti current collector (1st cycle). (b) Reversibility of Li+ intercalation/de-intercalation of electrode materials in 20 m LiTFSI. Cyclic voltammograms performed at 1 mV/s on Mo6S8 (dark blue), LFP (light blue), LTP (yellow), NMC622 (orange) as WE, YP50 activated carbon as CE and Ag/AgCl as reference electrode. All experiments were performed at room temperature (RT).

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 63

Cycling performances in aqueous superconcentrated electrolyte on the negative electrode side: role of concentration, cycling rate and temperature

In this section, we will focus on the cycling performances of a Mo6S8/LFP cell using

either 20 m LiTFSI or 20 m LiTFSI : 8 m LiBETI to assess the viability of these aqueous

superconcentrated electrolytes in practice. Cell performances were estimated with

electrochemical tests upon varying concentrations, cycling rate (C-rate defined by 1C

corresponding to one Li+ inserted in one hour) and temperatures.

1.1.1 Influence of concentration and C-rate

The capacities in charge and discharge of a Mo6S8/LFP cell were measured as a

function of the salt concentration from 5 m up to 20 m, corresponding to Water-in-salt

electrolyte. Cell capacity and Coulombic efficiency measured at 1C at room temperature

are shown in Figure 2. 3a and b. The difference between charge and discharge capacities

is becoming greater when lowering the salt concentration, i.e., the Coulombic efficiency

is decreasing. Furthermore, the capacity is found to fade over cycling much faster when

lowering the salt concentration. These results can tentatively be interpreted either as

the sign that no SEI is formed at lower concentrations or by invoking a greater solubility

of inorganic compounds forming the SEI, such as LiF134,146, Li2O146 or LiOH134 previously

observed forming on the surface of negative electrodes, at lower concentrations. Either

way, the continuous parasitic reactions occurring on the surface of Mo6S8 negative

electrode consume Li+ and cause the performances to decay over cycling at low

concentration. Finally, and more interestingly, the initial capacity in charge is found, in

Figure 2. 3a, similar for the different concentrations: 125 mA.h/g at 5 m, 120 mA.h/g at

10 m and 123 mA.h/g at 20 m, suggesting that the nature and the number of parasitic

reactions are independent on the concentration during the first charge, before the

formation of a SEI.

64 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Figure 2. 3 Effect of concentration for Mo6S8/LFP full cells measured in LiTFSI-based aqueous electrolytes. (a) Capacities of charge and discharge and (b) Coulombic efficiency (CE) as function of cycle number for several concentrations: 5 m LiTFSI (brown square), 10 m LiTFSI (purple square), 20 m LiTFSI (yellow square). Constant current measurements were performed at 1C at room temperature (RT).

Even though the cycling performances improve with concentration, they are

dependent on the C-rate, as shown in Figure 2. 4a and b where the evolution of the

charge capacity (Figure 2. 4a) and the Coulombic efficiency (Figure 2. 4b) are reported

for several C-rates at room temperature. Indeed, it clearly appears that the faster the

cycling rate is, the higher the Coulombic efficiency. However, the absolute value for the

charge capacity is slightly lower, hence leading to significant improvements in the cell

capacity retention. This phenomenon can be related to greater amount of parasitic

reactions occurring as time spent at potential close to the HER potential increases when

lowering the C-rate, thus “artificially” increasing the charge capacity at the expense of

the discharge capacity. As a result, and as often seen in the literature153, one obvious

way to increase the Coulombic efficiency and cycling performances of such systems is

by increasing the C-rate. However, our work reveals that C-rate below 1C must be

employed to accurately evaluate the performances of aqueous systems in this

configuration.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 65

Figure 2. 4 Effect of C-rate for Mo6S8/LFP full cells measured in LiTFSI-based aqueous electrolytes. Capacity of charge (a) and Coulombic efficiency (b) as function of cycle number for several C-rate: 0.15C, 0.25C, 0.5C, 1C, 2C, 4.5C. Constant current measurements were performed in 20 m LiTFSI electrolyte at room temperature.

1.1.2 Influence of temperature

Figure 2. 5 reveals that the effect of temperature is more pronounced at low C-rate

than at C-rate above 1C. Indeed, an operating temperature of 55 °C leads to a rapid

decay of the reversible capacity and a drastic drop of the Coulombic efficiency, both

leading to a shorter lifetime for the battery at low C-rate. Moreover, we can observe at

55 °C a larger charge capacity at the beginning of cycling associated with lower

Coulombic efficiency (see Figure 2. 5a and Figure 2. 5b), demonstrating a greater amount

of parasitic reactions at higher temperatures. Besides the enhancement of the global

degradation of the cell (loss of electrical contact, faster aging of materials, etc.) at high

temperatures, the origin for this phenomenon can be either kinetics or

thermodynamics. On the kinetics side, a higher temperature will both enhance the HER

kinetics, as well as the SEI degradation rate for SEI-components such as LiF, LiOH or Li2O

as reported in literature134,146. On the thermodynamics side, note that the HER potential

shifts towards lower potential by 160 mV between 25 °C and 55 °C (EHER @25 °C = 2.16 V

vs Li+/Li and EHER @55 °C = 2 V vs Li+/Li), while the Li insertion potential for Mo6S8 only shifts

by 20 mV (see Figure 2. 5b). This difference leads to the appearance of a plateau

attributed to the HER (as proved by gas monitoring) before the cell potential reaches

the cut-off of 1.5 V, again artificially increasing the charge capacity of the LIB while

reducing the lifetime.

66 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Figure 2. 5 Temperature impact for Mo6S8/LFP full cells measured in 20 m LiTFSI. (a) Charge and discharge capacities as a function of cycle number at 55 °C for several C-rates: 0.15C, 0.25C, 0.5C, 1C, 2C. (b) Potential as function of time for cells cycled at 0.5C at 25 °C and at 55 °C showing the shift of the HER plateau depending on the temperature. (c) Comparison between discharge capacities and Coulombic efficiency as function of cycle number for cells cycled at 0.5C at 25 °C (yellow) and 55 °C (orange).

1.1.3 Reproducibility issue

During the temperature study using the 20 m LiTFSI electrolyte, we could observe for

some cells cycled at 0.5C at 55 °C a large dispersion of charge and discharge capacities

values (see Figure 2. 6a). Our results at 55 °C (capacity vs cycle number) for three

different cells show charge capacity during the first charge ranging from 107 mA.h/g to

142 mA.h/g, far greater than the range for charge capacities determined at room

temperature (112 mA.h/g < Qcharge @RT < 123 mA.h/g) (see Figure 2. 6a). This

phenomenon is rooted in the effect of temperature that exacerbates small variations in

the SEI formation and stability, which in turn leads to different microstructures

(thickness and density) and thus damaging rate when cycled at 55 °C. Such variation is

highlighted in Figure 2. 6b with the appearance of a HER plateau at 55 °C before the cell

cut-off.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 67

Figure 2. 6 (a) Charge and discharge capacities as function of cycle number for three Mo6S8/LFP cells cycled at 0.5C at 55 °C and range of values for three cells cycled at 0.5C at room temperature that fall with the shaded grey area. (b) Potential as a function of time for three cells cycled in similar conditions, at 0.5C and 55 °C, showing the poor reproducibility of cell performances at elevated temperature.

To further study the viability of aqueous superconcentrated electrolytes, 20 m LiTFSI :

8 m LiBETI Water-in-bisalt electrolyte was then investigated. Indeed, such electrolyte, in

which the water amount is even smaller (salt to water molar ratio is set to 1:2 in 20 m

LiTFSI : 8 m LiBETI compared to 1:2.8 in 20 m LiTFSI), was previously propose to provide

better cycling performances than classical 20 m LiTFSI electrolyte, thus enabling to cycle

low potential negative electrode such as LTO58.

1.2.1 Influence of the C-rate: 0.15C vs 1C at RT and temperature: RT vs 55 °C at 0.15C

Based on the results shown in section 2.1.1, the C-rate study was limited to C-rates

no greater than 1C to accurately evaluate the performances of WiBS. Therefore, two C-

rates of 1C and 0.15C were chosen to illustrate WiBS behavior with cycling rate. Figure

2. 7a shows the evolution of the charge capacity and the Coulombic efficiency with

cycles. The influence of the C-rate for WiBS-based cell performances is in line with the

results previously obtained in WiSE. Indeed, the faster the cycling rate is, the greater the

Coulombic efficiency, the lower the charge capacity. Cycling performances were then

tested at 55 °C (see Figure 2. 7b). Doing so, an increase of temperature was found to

lead to greater capacity in charge (132 mA.h/g at 55 °C compared to 119 mA.h/g at RT)

associated with lower Coulombic efficiency, in agreement with an increased amount of

parasitic reactions occurring at high temperature due to faster HER kinetics, greater HER

68 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

onset potential and faster SEI degradation rate combined with an enhancement of the

global degradation of the cell (loss of electrical contact, faster aging of materials, etc.),

alike previously observed in WiSE.

Figure 2. 7 Effect of C-rate and temperature for Mo6S8/LFP full cells cycled in 20 m LiTFSI : 8 m LiBETI LIB. (a) Capacity of charge and Coulombic efficiency as function of cycle number for two representative C-rates: 0.15C (dark blue) and 1C (light blue). Constant current measurements were performed in 20 m LiTFSI electrolyte at room temperature. (b) Temperature effect on WiBS-based LIB. Comparison between charge capacities and Coulombic efficiency as function of cycle number for cells cycled at 0.15C at room temperature (yellow) and 55°C (red).

1.2.2 Comparison with WiSE-based electrolyte at 55 °C at 0.15C

While all these measurements are very much in line with those previously obtained

for WiSE, the capacity fading is nevertheless found to be much slower when using WiBS

electrolyte than with WiSE (see Figure 2. 8), leading to longer shelf-life. Bearing in mind

that the concentration of water in WiBS is 1.6 times smaller than in WiSE, this

observation could at first be explained by a decrease of the number of water molecules

available for the HER and a decrease of the dissolution rate of the SEI in WiBS148,211.

However, this explanation is contradicted by recent studies showing that water

molecules reduced in the HER process are those solvating lithium cation212 and that the

first solvation sheath of lithium cation is rather similar in WiSE and WiBS, therefore the

reactivity of water in these two electrolytes should be alike. A more likely possibility is

the viscosity difference between both electrolytes that is about 6 times greater for WiBS

(203 mPa.s at 30°C58) than for WiSE electrolyte (36.2 mPa.s at 25°C111). Indeed, a greater

viscosity would limit the HER kinetics and the degradation rate of the SEI, thus enabling

better performances.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 69

Figure 2. 8 Comparison of WiSE and WiBS performances with cycling. Charge and discharge capacity as function of cycle number for Mo6S8/LFP full cells cycled at 0.15C at 55 °C.

Altogether, studies varying concentration, C-rate and temperature show that 20 m

LiTFSI-based LIB still suffer from damaging parasitic reactions, which will be discussed in

the following section dedicated to the water reduction reaction.

Origin of the performances decay: a gas monitoring study

Having demonstrated the effect of cycling conditions on parasitic reactions, we then

performed operando gas monitoring using a combination of online electrochemical

mass spectrometry (OEMS) and pressure cells (i) to highlight that water reduction takes

place during cycling and (ii) to qualitatively and (iii) quantitatively interrogate the impact

of gas evolution on the battery performances.

70 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Figure 2. 9a shows the potential and pressure changes as a function of time during

cycling for 20 m LiTFSI electrolyte. The cell pressure is found to increase when the

potential reaches the 2nd lithium insertion plateau of Mo6S8 at 2.3 V vs Li+/Li (≈ 1.4 V in

a complete Mo6S8/LFP cell, see Figure 2. 9b for the definition of the plateau). Strikingly,

the pressure never stops increasing in this configuration during cycling, demonstrating

that parasitic reactions keep occurring, consistent with the low Coulombic efficiency

observed in Figure 2. 4b. Moreover, from OEMS measurements (Figure 2. 9b), we could

deduce the formation of gaseous hydrogen as soon as Mo6S8 reaches its second lithium

insertion plateau. Therefore, water reduction producing hydrogen is responsible for the

pressure increase, and any SEI formed on the electrode at these potentials is not

protective enough to prevent the continuous consumption of WiSE during cycling.

Besides, it is important to notice that hydrogen evolution competes with lithium

insertion, but the former does not prevent the latter. Hence, two rates for the

electrochemical hydrogen evolution were observed in Figure 2. 9b. The first rate starts

concomitantly with the 2nd insertion of lithium around 1.38 V. However, this first rate is

slow compared to the one kicking in once the electrode is fully lithiated, when the

potential goes above 1.4 V and where all electrons are consumed toward the HER.

Moreover, as seen in Table 2. 1, the discharge capacity recorded with a pressure cell

cycled at 0.10C remains stable at ≈ 107.5 mA.h/g during the four first cycles, unlike the

charge capacity which is always greater and varies from 126 to 117 mA.h/g. This result

indicates that lithium insertion into Mo6S8 is not affected by the HER during charge, as

the cell provides the same discharge capacity over the first cycles. However, the

continuous consumption of water via the HER may eventually lead to the crystallization

of the salt and ultimately the drying out of the cell, that will be prejudicial for practical

application153.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 71

Figure 2. 9 Gas monitoring for a Mo6S8/LFP full cell measured in 20 m LiTFSI. (a) Potential (black line) and pressure (red line) as function of time at 0.10C monitored in a pressure cell. (b) Potential (black line) and hydrogen evolution (red line) as function of time at 0.15C monitored using an online electrochemical mass spectrometry (OEMS) cell. Experiments performed at 25 °C.

Table 2. 1 Discharge and charge capacities and Coulombic efficiency for a Mo6S8/LFP pressure cell cycled at 0.10C and 25 °C in 20 m LiTFSI.

Figure 2. 10 shows the hydrogen evolution recorded during cycling at 0.15C and 55°C

by OEMS. First, these results indicate that, as already observed at 25 °C, hydrogen

production starts on the 2nd Li insertion plateau at 1.37 V at 55 °C. Then, the hydrogen

production is greater as the temperature increase, due to the enhancement of both the

HER kinetics and SEI degradation rate. Indeed, an approximate linear fit of the first slope

-corresponding to hydrogen production during 2nd insertion plateau- is four times

greater at 55 °C than at 25 °C (≈ 3.4 ∙ 10-11 uma/h at 55 °C compared to ≈ 0.81 ∙ 10-11

uma/h at 25 °C). Eventually, as shown in Figure 2. 5, increasing the temperature to 55

°C may lead to the apparition of a 3rd plateau around 1.46 V that can be attributed to

pure HER.

Cycle number Discharge capacity

[mA.h/g] Charge capacity [mA.h/g] CE [%]

1 107.66 126.25 85.2

2 107.62 119.88 89.8

3 107.47 118.53 90.7

4 107.44 117.84 91.2

72 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Figure 2. 10 Online electrochemical mass spectrometry (OEMS) performed in 20 m LiTFSI for a Mo6S8/LFP full cell at 55 °C. Potential (black line) and hydrogen evolution (red line) as function of time at 0.15C.

The potential and the pressure changes plotted as function of time during cycling in

20 m LiTFSI : 8 m LiBETI (see Figure 2. 11a) show that the cell pressure continuously

increases during cycling, as the result of the evolution of H2 spotted by OEMS

measurement (see Figure 2. 11b). Hence, alike for WiSE electrolyte, continuous water

consumption occurs in parallel with lithium insertion for WiBS electrolyte.

Figure 2. 11 Gas monitoring for a Mo6S8/LFP cell measured in 20 m LiTFSI: 8 m LiBETI WiBS. (a) Potential (black line) and pressure (red line) as function of time at 0.10C monitored using a pressure cell. (b) Potential (black line) and hydrogen (red line)

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 73

evolution as function of time at 0.10C monitored using an online electrochemical mass spectrometry OEMS cell.

When comparing the relative pressure increase for two pressure cells during cycling

in 20 m LiTFSI and 20 m LiTFSI : 8 m LiBETI (see Figure 2. 12), one can observe that alike

for cycling performances, WiBS-based battery gives better results, i.e., less gas

production. However, the use of WiBS does not prevent water reduction to occur.

Figure 2. 12 Comparison of gas evolution and cycling performances at high temperature in Mo6S8/LFP cells using either WiSE or WiBS. (a) Potential (black line) and relative pressure increase (purple line) as function of time at 0.10C monitored using a pressure cell in 20 m LiTFSI electrolyte. (b) Potential (black line) and relative pressure increase (red line) as function of time at 0.10C monitored using a pressure cell in 20 m LiTFSI : 8 m LiBETI electrolyte.

To evaluate further the importance of hydrogen gas release in the total parasitic

reactions, we plotted its amount as a function of irreversible capacity per cycle for

several C-rates (see Figure 2. 13) and note a nearly linear trend with, at 0.1C, an

irreversible capacity of 12 mA.h/g that corresponds mainly to a gas release of 1.4 µmol

in addition to the contribution of other side reactions.

74 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Figure 2. 13 Amount of gas released per cycle as function of irreversible capacity (dash line is a guide to the eyes). Experiments performed at room temperature using Mo6S8/LFP full cells in 20 m LiTFSI WiSE.

Such an hydrogen evolution originates from the decomposition of H2O that could

proceed either via a direct or indirect process according to Equation 2. 1 and Equation

2. 2, respectively.

Equation 2. 1: Direct HER

2 ∙ 𝐻2𝑂 + 2 ∙ 𝑒− → 2 ∙ 𝐻𝑂− + 𝐻2

Equation 2. 2: Indirect HER

𝐿𝑖4𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐻2𝑂 → 𝐿𝑖4−𝑥𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐿𝑖+ + 𝑥 ∙ 𝐻𝑂− + 𝑥

2∙ 𝐻2

On the basis of the direct mechanism (Equation 2. 1), we can from simple calculations

(based on the ideal gas law) nearly account for the amount of H2 released either during

pressure cells or OEMS experiments (see Table 2. 2 and Table 2. 3 for listed results and

Figure M.M. 11, Figure M.M. 12 and Figure M.M. 13, in Chapter Materials and Methods

for calculation details). The results implies that Equation 2. 1 is by far majority in the

total irreversible capacity (70 % of the irreversible capacity per cycle) regardless the use

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 75

of WiSE or WiBS and the C-rate, suggesting that the nature of parasitic reactions taking

place in WiBS electrolyte is identical to the one in WiSE. Therefore, such a system will

certainly face similar issues to WiSE electrolyte, with nevertheless a decreased amount

of hydrogen evolution (as shown in Figure 2. 12).

Table 2. 2 Coulombic efficiency and gas quantification by operando pressure cells (average value on the first five cycles) as function of C-rate for a Mo6S8/LFP full cell cycled in WiSE or WiBS electrolyte.

Electrolyte C-rate 1st Coulombic efficiency [%] 𝑄𝐻𝐸𝑅

𝑄𝑖𝑟𝑟 [%]

WiSE 0.10C 89.2 67

WiSE 0.15C 92.7 78

WiSE 0.25C 95.8 79

WiSE 1C 98 67

WiBS 0.10C 90.2 71

Table 2. 3 Coulombic efficiency and gas quantification by OEMS (1st cycle) for a Mo6S8/LFP cell cycled in WiSE or WiBS electrolyte.

Electrolyte C-rate Coulombic

efficiency [%]

𝑄𝐻𝐸𝑅

𝑄𝑖𝑟𝑟 [%]

WiSE 0.15C 73.2 79.4

WiBS 0.10C 79.9 86.9

As a conclusion, the study of the impact of concentration, cycling rate and

temperature on the cell performances combined with operando gas monitoring

highlight that water reduction is not prevented during cycling in aqueous

superconcentrated electrolytes. To further explore the practical viability of WiSE and

WiBS, the following section will focus on self-discharge protocol assessing the cell

performances during resting period.

76 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Self-discharge protocol to assess aqueous superconcentrated electrolytes viability during resting period

The impact of self-discharge on cell performances was estimated using coin cells. Two

cycling protocols were used to assess the cell open circuit voltage (OCV) decay during

self-discharge and thus the SEI stability. The first one, depicted in Figure 2. 14a,

corresponds to a single full charge followed by an OCV period until the cell voltage gets

to 1.31 V (2.34 V vs Li+/Li) where the 2nd plateau is entirely over, i.e. the cell state-of-

charge reaches 25 %. Indeed, as mentioned in the introduction of the chapter, the 2nd

insertion plateau -corresponding to a cell voltage of 1.32 V in discharge- accounts for

three Li+ cations over the four that Mo6S8 can insert/de-insert. Moreover, to be able to

extend the SEI formation time during the first few charges, a second protocol employing

a pre-cycling step consisting of five consecutive charge/discharge cycles followed by one

last charge was used, as displayed in Figure 2. 14b. Alike for the single-charge protocol,

the cell was then discharge at 25 % SOC and several C-rates, from 0.15C to 4.5C, were

tested to vary the SEI formation time.

Figure 2. 14 Illustration of the self-discharge protocols performed in this study. (a) Open circuit voltage (OCV) decay measured upon rest for a cell fully charged at 0.5C. (b) OCV decay measured for a fully charged cell after five pre-cycles at 0.5C at room temperature.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 77

In Figure 2. 15 is compared the time needed to reach 25 % state-of-charge (SOC)

during self-discharge with the time spent to form the SEI during charge. As discussed in

Chapter 1, the SEI formation either through the direct reduction of TFSI- anions or their

chemical reaction with hydroxide anions produced by HER occurs at cell voltage above

1.3 V (i.e. 2.34 V vs Li+/Li at the negative electrode). Thus, the stability of the SEI must

be dependent on the cycling rate and/or the number of consecutive charges, both

defining its thickness and density. Indeed, one can notice in Figure 2. 15 that the longer

the time spent to form the SEI, the slowest the self-discharge. Moreover, when more

than 20 h are used to form the SEI, the time needed to discharge the cell to 25 % SOC

remains roughly stable (≈ 900 h needed) and no further improvement in the SEI stability

is observed. However, we must acknowledge that reproducibility tests were not

performed. Furthermore, the number of parameters tested is far from being exhaustive.

Therefore, an in-depth study would be needed to draw a solid trend regarding the SEI

stability and identify and optimize the key parameters (absolute time spent on charge

or repeated cycles, temperature etc.) controlling the SEI formation step in aqueous

superconcentrated electrolytes.

Figure 2. 15 Assessment of the SEI stability. Time needed to reach 25 % state-of-charge by resting period as function of the time spent to form SEI during charge at room temperature. Mo6S8/LFP full cells using 20 m LiTFSI were used.

78 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

3.2.1 Long cycling impact on cell performances at room temperature

To assess the battery performances when the cell faces self-discharge period, long

cycling with self-discharge protocol was performed. 0.5C C-rate was chosen as a

compromise between fast experiments to quickly assess the cell performances and

sufficiently slow C-rate to highlight parasitic reactions.

In Figure 2. 16a is shown the protocol, called 20 h OCV protocol, using a 20 h OCV

step at the end of each charge after one cycle of charge/discharge/charge. The second

protocol, called 10 cycles -10 h OCV protocol, (see Figure 2. 16b) consists of a pre-cycling

step of 10 cycles to form the SEI. After that, the cell is let to rest for 10 h every 10 cycles.

Cycling with or without self-discharge step are compared in terms of discharge capacity,

normalized discharge capacity and Coulombic efficiency.

Figure 2. 16 Illustration of the self-discharge protocols performed in this study. (a) 20 h OCV protocol using a 20 h OCV period after each charge. (b) 10 cycles-10 h OCV protocol using a 10 h OCV period after 10 cycles of charge/discharge. All Mo6S8/LFP cells were cycled at 0.5C at room temperature in 20 m LiTFSI.

Clear conclusions can be drawn comparing results obtained with a continuous cycling

protocol, the 20 h OCV protocol and the 10 cycles-10 h OCV protocol. Figure 2. 17 shows

the mean values and the standard deviation (over 3 cells) of discharge capacity,

normalized discharge capacity by the first discharge capacity and the Coulombic

efficiency for each protocol.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 79

Figure 2. 17 Impact of the self-discharge protocol on the cell performances. Continuous cycling (red), 20 h OCV protocol (light purple), 10 cycles-10 h OCV (dark purple). (a) Mean discharge capacity obtained at 0.5C at room temperature as function of cycle number. (b) Mean normalized discharge capacity by the first discharge capacity obtained as function of cycle number. (c) Mean Coulombic efficiency as function of cycle number. All Mo6S8/LFP cells were cycled at 0.5C at room temperature in 20 m LiTFSI. Three cells were used for each protocol.

80 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

First looking at the discharge capacities (see Figure 2. 17a), a drop of capacity of about

4 mA.h/g is observed for both self-discharge protocols after resting period. Thus, the

use of a resting period leads to a drop of discharge capacity of 4 % (see Figure 2. 17b)

compared to a continuous cycling, regardless the resting protocol. Moreover, for the

20h OCV protocol, after resting both the Coulombic efficiency and the discharge capacity

drop from the first cycle and stabilize at around 96 % and 100 mA.h/g, respectively (see

Figure 2. 17a and c). Hence, while the parasitic reactions occurring during the first charge

are similar (the Coulombic efficiencies being around 95 % during the first cycle for both),

when performing the 20 h OCV protocol, the cell degradation is enhanced as a result of

the extended period spent at OCV. These observations are further confirmed with the

10 cycles-10 h OCV protocol where each resting period is followed by a decrease of

capacity in discharge, normalized capacity of discharge and Coulombic efficiency close

from the values obtained with the 20 h OCV protocol. Altogether, these data suggest a

partial degradation of the SEI or a partially porous SEI. Thus, the SEI cannot fully prevent

the HER, therefore explaining that 98 % Coulombic efficiency cannot be passed with this

system.

Besides, the coupling of self-discharge protocol (as described in Figure 2. 16a) with

pressure cell experiments enable to show that water reduction also happen during

resting period.

Figure 2. 18 Pressure cell assessing gas evolution during 20 h OCV protocol illustrating self-discharge. (a) Potential (black line) and pressure (red line) as function of time for a Mo6S8/LFP cell cycled at 0.5C and 25 °C. (b) Zoom of the pressure evolution (red line) as function of time during resting period.

As shown in Figure 2. 18, when a cell is stored at 100 % SOC, pressure increase takes

place during resting. As described in Chapter Materials and Methods, the application of

the ideal gas law enables to determine a water consumption of 0.13 µmolwater/hself-

discharge (mean value for two pressure cells, cycled at 0.5C, 25 °C with a similar mass of

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 81

active material). Assuming a two electrons reaction for water reduction, one can

calculate the capacity related to such water consumption and compare it to the

discharge capacity loss due to 20h OCV period (as described in Chapter Materials and

Methods). Thus, the proportion of water reduction related to self-discharge is found to

be between 75 % and 100 % of the discharge capacity loss for two pressure cells cycled

in the same conditions.

Nonetheless, these observations contrast with a report published in 2016 by Suo et

al.213 in which self-discharge was assessed using the 10 cycles-10 h OCV protocol at 0.5C.

There, a steady increase of the Coulombic efficiency following each of the 10 hours

resting period was observed, as reproduced in Table 2. 4. First, when reproducing this

protocol, Coulombic efficiency obtained in this work was found to stabilize after 3-4

cycles at values of 98 %, while 30 cycles were required to reach the same Coulombic

efficiency as reported by Suo et al.213. Bearing in mind that cycling protocols are

identical, the SEI stability is expected to be similar. Such differences thus most likely

arise from the morphology of the electrode materials and/or the cell assembly (ratio

mass of active material vs volume of electrolyte, etc.). Second, we found a constant loss

of capacity of about 4 mA.h/g upon rest at room temperature while this capacity loss,

initially of 6 mA.h/g after the first resting period, decreases on the subsequent resting

periods in Suo’s report. In conclusion, our results evidence a continuous loss of capacity

upon subsequent resting highlighted by the discharge capacity that smoothly decays

upon cycling.

Table 2. 4 Comparison of the Coulombic efficiencies and discharge capacities obtained after each 10h resting period at OCV in Ref213 with those experimentally obtained in this work.

Cycle number (following the

10h OCV period)

CE as reported by Suo213 [%]

CE (obtained in this work) [%]

Discharge Capacity (obtained in this work) [mA.h/g]

0 82 94.3 103 ± 2

11th 80.6 95.7 98.9 ± 1.5

21th 84.7 95.8 98.5 ± 1.5

31th 87.6 95.9 97.9 ± 1.6

41th 89.1 95.9 97.2 ± 1.7

Importantly, one can notice that the self-discharge is reversible in terms of lithium

balance. Indeed, both the Coulombic efficiency and the discharge capacity are recovered

upon cycling after a resting time. Indeed, by comparing the discharge capacity obtained

82 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

before and after the 10h OCV period, one can find that discharge capacity recovers from

99.7 % to 100 % even after 13 repetition of the 10 cycles-10 h OCV protocol.

Figure 2. 19 Reversibility of Lithium loss during self-discharge. (a) Illustration of protocol reaching 100 % self-discharge for a Mo6S8/LFP cell fully charged at 0.5C followed by continuous cycling at 0.5C. (b) Charge and discharge capacities as function of cycle number (mean value over three cells for continuous cycling protocol), (c) Coulombic efficiency as function of cycle number capacity obtained for three cells continuously cycled and one cell which endured 100 % self-discharge after the first charge. All Mo6S8/LFP cells were cycled at 0.5C at room temperature in 20 m LiTFSI.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 83

To confirm that the self-discharge is mostly reversible in terms of Li balance, a unique

charge was performed at 0.5C followed by a complete self-discharge. Once the cell

reaches 0 % SOC, continuous cycling was done at 0.5C. First of all, one can observe in

Figure 2. 19a that the self-discharge process follows the electrochemical signature of

Mo6S8. However, the first plateau (at 0.97 V i.e. 2.7 V vs Li+/Li) is longer than the second

one (at 1.33 V i.e. 2.3 V vs Li+/Li), ≈1400 h and 900 h, respectively. This is in contrast with

the fact that the first plateau accounts for ¼ and the second plateau for ¾ of the Li

insertion/de-insertion. Indeed, during self-discharge, the removal of Li from the

electrode material depends on the kinetics of the reaction (Equation 2. 2), thus the

length of the plateaus may be modified if the water reduction kinetics is different for

each plateau, which will be discussed in greater details in Chapter 3. However, one can

already note that self-discharge is relatively slow since more than three months are

needed to fully self-discharge the cell. Regarding the charge and discharge capacities,

Figure 2. 19b and c compare the performances for 3 cells cycled continuously at 0.5C

with that of a cell that first underwent a 100 % self-discharge after the first charge. Doing

so, one can first observe that the cell which endured a 100 % self-discharge exhibit lower

charge and discharge capacity, the difference being however limited. Moreover, the

Coulombic efficiency is found to stabilize to similar values (between 0.98 and 0.99) after

10 cycles for both protocols (Figure 2. 19c). This confirms that the Li loss underwent

during self-discharge is reversible.

3.2.2 Temperature effect on self-discharge

Acknowledging that temperature will enhance the SEI instability, 55 °C cycling tests

were performed with the 10 cycles-10 h OCV protocol where SEI is expected to be stable.

In Figure 2. 20 is plotted the discharge capacity and Coulombic efficiency as function of

cycles for experiments carried out at room temperature and 55 °C. First, one can observe

that applying 10 h resting period leads to a similar behavior regardless of the

temperature. Indeed, both discharge capacity and Coulombic efficiency fall after the

OCV period. However, as shown in at 0.5C.

Table 2. 5, the drop in capacity is enhanced at 55 °C. This result is consistent with a

faster SEI degradation and an enhanced HER kinetics at 55 °C, both leading to an

increased drop in cell performances.

84 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

Figure 2. 20 Impact of temperature on self-discharge performances using a 10 cycles-10 h OCV protocol. Discharge capacity and Coulombic efficiency as function of cycle number for Mo6S8/LFP full cells cycled at room temperature (yellow) and 55 °C (red) at 0.5C.

Table 2. 5 Capacity loss after 20h OCV resting period for Mo6S8/LFP cells. Comparison at 25 °C and 55 °C.

Having investigating in depth the 20 m LiTFSI WiSE system, our attention then turned

to the stability of the SEI formed in WiBS 20 m LiTFSI : 8 m LiBETI. For that, the same

methodology as previously used for WiSE was employed and similar behavior was

observed for the 20 h OCV protocol. Hence, the loss in discharge capacity observed after

a 20 hours OCV period applied after a first cycle performed at 1C is found identical to

the discharge capacity loss measured in WiSE for the same C-rate, i.e. ≈ 9 mA.h/g.

Temperature RT 55°C

Capacity loss [mA.h/g] 4 16

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 85

Figure 2. 21 Impact of the 20 h OCV protocol on cell performances using WiBS electrolyte. Charge and discharge capacity as function of cycle number for continuous cycling (red) and 20 h OCV protocol (light purple) for Mo6S8/LFP full cells cycled at 1C at room temperature.

As a conclusion, the use of self-discharge protocols enabled us to highlight the

instability and/or the lack of protective power of the SEI formed in aqueous

superconcentrated electrolytes. Therefore, while optimizing the formatting conditions

can help delaying the degradation of the SEI, as intensively experimented in non-

aqueous Li-ion batteries, this certainly will not prevent the drying out of the cell over

prolonged time, especially at temperature greater than room temperature as it will be

discussed in Chapter 3.

Cycling viability on the positive side: a gas monitoring study

Besides the SEI forming at the negative electrode, another important aspect in

selecting electrolytes regards their stability at the positive electrode under highly

oxidizing potentials. Having established that the cycling performances for WiSE-based

86 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

aqueous batteries will certainly be limited by the absence of passivating SEI at the

negative electrode, we focused our attention on gassing experiments rather than on

cycling performances for the positive electrode.

Figure 2. 22a shows the evolution of pressure and potential as a function of time for

a LTP/NMC622 cell using WiSE electrolyte at 25 °C. There, even when pushing the

potential cut-off up to 2 V (4.78 V vs Li+/Li for NMC622), any pressure increase can hardly

be detected, with only a pressure increase of 0.6 mbar (0.22 µmol of gas) being

observed, this amount being within the detection limit of this technique. This absence

of gassing is consistent with the previously reported formation of a TFSI-rich double

layer, preventing water to access the interface, as described in Chapter 1. Furthermore,

this observation is also consistent with the OER kinetics being very sluggish when

compared to the HER, as widely discussed in the electrocatalysis field211. Hence, at 25

°C, WiSE electrolyte seems to be stable and not to face any drastic degradation, in

agreement with the electrochemical stability of superconcentrated aqueous

electrolytes under anodic polarization reported in previous studies58,214.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 87

Figure 2. 22 Potential (black line) and pressure (orange line) measured as a function of time for NMC622/LTP full cells cycled in (a) Water-in-salt 20 m LiTFSI electrolyte at 25 °C, (b) at 55 °C (*note that the peak observed below 80 hours for the pressure is due to an opening of the oven) and (c) Water-in-bisalt 20 m LiTFSI : 8 m LiBETI at 55 °C.

Following these measurements at 25 °C, the anodic stability of WiSE was assessed at

higher temperature by cycling pressure cells at 55 °C. The evolution of pressure and

potential as a function of time are reported in Figure 2. 22b. Compared to the results

88 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

obtained at room temperature, a significant gas evolution could be spotted during

delithiation/oxidation of NMC622 with two slopes being observed. The first one that

leads to a pressure increase of 4.3 mbar (1.59 µmol) occurs between 1.1 V and 1.4 V,

and can be explained by the competition between the electrochemical Li+ de-

intercalation and the slow parasitic reactions, either direct (OER) or indirect (self-

discharge)215. The second one starting above 1.4 V (4.2 V vs Li+/Li) leads to a greater

production of gases of ≈ 7 mbar (2.59 µmol) and can be mainly attributed to parasitic

reactions, which can also be responsible for the appearance of a plateau at high

potential (≈ 2 V) which is solely present during the first charge. Indeed, NMC is known

to face greater degradation at high temperatures and high potential cut-off216–218.

However, the origin of the gas production certainly arises from the corrosion of the

carbon additive at high potential in aqueous media, as spotted by OEMS measurements

during which CO2 is detected (see Figure 2. 23)219,220. Nonetheless, the detection of more

than one gas, during these measurements, prevents us from performing quantification

to determine the amount of mole produced by each gases. For sake of completion, the

stability of WiBS-based electrolyte was also tested using pressure cells at 55 °C (see

Figure 2. 22c), and similar behavior is observed as for WiSE. Hence, the pressure increase

recorded during the de-insertion plateau between 1.1 V and 1.4 V is ≈ 3 mbar (1.11

µmol) in WiBS, compared to 4 mbar in WiSE. This first gas release is followed by an

additional pressure increase of 8 mbar (2.96 µmol) at greater potential, compared to 7

mbar previously measured for WiSE. Differences in pressure variations were considered

to be within the same ranges. To conclude, at elevated temperature, the stability of the

NMC622 self-standing electrode/superconcentrated aqueous electrolyte assembly is

compromised under anodic polarization when compared to room temperature.

Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB 89

Figure 2. 23 Online electrochemical mass spectrometry performed on NMC622/LTP full cell using WiSE electrolyte. Potential (black line) and gas evolution (yellow line) as function of time are plotted for carbon dioxide (CO2: m/z=44). Experiment performed at 55 °C, at C/40. Note the gap is attributed to a partial stop of the experiment.

Conclusion of the chapter

In this chapter, an in-depth study of the cathodic and anodic stability of

superconcentrated aqueous electrolytes as function of the operating conditions was

carried out. First, the assessment of the capacity retention and Coulombic efficiency

with increased LiTFSI concentration shows the benefic influence of superconcentration

on cycling performances. However, such improvement was quickly nuanced when

performing tests at different C-rate. Indeed, when cycled at low C-rate (below 1C), the

damaging effect of parasitic reactions such as hydrogen evolution reaction are brought

to light. Their detrimental impact is exacerbated at higher temperature. To determine

the origin of the performances decay, we then performed operando gas monitoring.

Combining pressure cell tests with online electrochemical mass spectrometry, hydrogen

production was detected to start concomitantly with Li insertion at a cell voltage of ≈

1.38 V (2.3 V vs Li+/Li at the negative electrode). This gas evolution never stops during

cycling, testifying of the poor SEI protective power against HER. Eventually, the practical

viability of WiSE electrolyte in LIB was assessed by self-discharge protocols. Several self-

90 Chapter 2 - Cycling viability of aqueous superconcentrated electrolytes in LIB

discharge protocols were thus applied to both assess the SEI stability as a function of its

formation time and the impact of self-discharge during long cycling. Altogether, these

results show that even though self-discharge is partially reversible in terms of lithium

balance, cell life is limited when self-discharge protocols are applied. Furthermore, to

assess the behavior of aqueous superconcentrated electrolyte in LIB, a non-exhaustive

but representative study was performed using 20 m LiTFSI : 8 m LiBETI WiBS electrolyte.

Even though the ESW is larger in WiBS electrolyte, HER was found to occur during

cycling, although less intense. Moreover, discharge capacity loss was observed after

resting period, showing the weak passivation of the SEI. Therefore, WiBS-based LIB faces

critical limitations identical to the ones encountered in WiSE. Finally, studying parasitic

reactions at the positive electrode side show limited evolution of oxygen and only at

elevated temperature damageable parasitic reactions were observed.

Altogether, the results presented in this chapter on the practical viability of aqueous

superconcentrated electrolyte on the negative electrode call for the design of a stable

SEI that will effectively passivate the negative electrode and prevent water reduction.

Indeed, the major drawback of aqueous superconcentrated electrolytes lays in the

continuous water consumption during both cycling and self-discharge, eventually

leading to the drying out of the cell. Therefore, Chapter 3 will first discuss the instability

of the SEI in WiSE. Then the irreversible consumption of water that may cause the

crystallization of the electrolyte, thus the drying out, will be presented. Finally, the

kinetics of water reduction during both self-discharge and cycling will be determined.

CHAPTER 3 – INSTABILITY OF NATIVE SEI LEADS TO

THE DRYING OUT OF AQUEOUS

SUPERCONCENTRATED LI-ION BATTERY

94 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

Introduction

As evidenced in Chapter 2, practical viability of Water-in-salt electrolyte (WiSE) is

limited by continuous parasitic reactions occurring at the negative electrode during

cycling and resting period. Online electrochemical mass spectrometry measurements

and pressure cells tests identified water reduction as the major parasitic reaction

limiting the cell viability. SEI instability and/or specific morphology may explain that

water reaches the electrode interface. Therefore, Chapter 3 will focus on the assessment

of the native SEI stability by electrochemical characterizations, cyclic voltammetry (CV)

and electrochemical impedance spectroscopy (EIS), before to focus on the effect of

water consumption on the cell life by using differential scanning calorimetry (DSC).

Besides, kinetics of water consumption through direct HER (2 ∙ 𝐻2𝑂 + 2 ∙ 𝑒− → 2 ∙

𝐻𝑂− + 𝐻2 ) or indirect self-discharge ( 𝐿𝑖4𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐻2𝑂 → 𝐿𝑖4−𝑥𝑀𝑜6𝑆8 + 𝑥 ∙

𝐿𝑖+ + 𝑥 ∙ 𝐻𝑂− + 𝑥

2∙ 𝐻2) mechanisms are assessed by electrochemical measurements

using CV, continuous cycling protocols and self-discharge tests as function of

temperature. Hence, based on the Arrhenius law, the activation energies of both

mechanisms can be extracted. Figure 3. 1 illustrates the outline of the chapter based on

the results discussed in Chapter 2.

Figure 3. 1 Scheme of the outline of Chapter 3 based on the results discussed in Chapter 2.

Eventually, the figure of merits for WiSE-based aqueous Li-ion batteries is provided

and compared with existing aqueous technologies (Lead-acid, Ni-Cd, Ni-MH) and aprotic

Li-ion batteries to assess the viability of this technology.

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 95

Probing the SEI instability in aqueous superconcentrated electrolytes

As shown in Chapter 2, the amount of irreversible capacity upon cycling associated

to HER (70 %) demonstrates that the SEI forming in WiSE is not passivating enough, nor

stable. To check the protective power of native SEI, SEI formation was mimicked by

cycling a glassy carbon working electrode in a 3-electrodes cell before applying a resting

period of one-hour and perform another CV measurement to measure the cathodic

current corresponding to the HER (see Figure 3. 2). Both 20 m LiTFSI and 20 m LiTFSI :

8 m LiBETI electrolytes were tested at 25 °C and 35 °C, respectively (see Figure 3. 2).

Moreover, the influence of temperature on the passivating abilities of aqueous

superconcentrated electrolyte was assessed using 20 m LiTFSI (see Figure 3. 3).

During the first CV scan, a peak at 0.8 V vs Li+/Li is observed in WiSE (see Figure 3. 2a),

and is attributed to HER on the surface of the glassy carbon working electrode134,149.

Upon cycling, the intensity of this peak decreases before to eventually almost vanish

after 15 cycles (see green line). This phenomenon is explained by the gradual passivation

of the glassy carbon electrode as a result of the SEI formation, as proposed by Dubouis

et al.134 and schematized in Figure 3. 2c. However, after applying a one-hour resting

period, the subsequent CV recorded (see purple line) attests that the passivation is lifted

as the peak intensity corresponding to the HER is back to that recorded during the very

first cycle. A similar behavior is found when performing the experiment in 20 m LiTFSI :

8 m LiBETI, although the reductive peak observed during first scan is shifted to lower

potential (0.5 V vs Li+/Li) (see Figure 3. 2b).

96 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

Figure 3. 2 Assessment of the SEI stability over time by mimicking its formation on inert material. (a) Cyclic voltammetry performed at 50 mV/s in 20 m LiTFSI in a 3-electrodes cell using glassy carbon as working electrode, Pt wire as counter electrode and silver wire as pseudo-reference. (b) Cyclic voltammetry performed at 35 °C (to avoid crystallization) at 50 mV/s in 20 m LiTFSI : 8 m LiBETI in a 3-electrodes cell using glassy carbon as working electrode, Pt wire as counter electrode and saturated calomel electrode as reference. (a-b) The first (red) and the fifteenth (green) voltammograms are shown, as well as the one recorded after one-hour open circuit voltage (OCV) (purple). (c) Illustration of the SEI formation and its partial degradation after a resting period of 1 h.

Besides, similar experiments were also performed as function of temperature (at

35 °C, 45 °C and 55 °C) in WiSE, as shown in Figure 3. 3, and similar trends were observed

with the passivation of glassy carbon. Thus, the passivating layer is deteriorated during

resting period, further confirming our self-discharge measurements discussed in

Chapter 2. The presence of two peaks in the cyclic voltammetry performed after the

resting period at 35 °C in WiBS (see Figure 3. 2b) and at 55 °C in WiSE (see Figure 3. 3d)

may be attributed to shifts in potential due to the presence of bubbles at the electrode

interface. Moreover, a competition between the precipitation of LiTFSI, as recently

proposed149, and the dissolution of LiF can contribute to this dynamic SEI behavior.

Indeed, bearing in mind that the ratio volume of the electrolyte/active material (several

mL vs 4 mm diameter of glassy carbon) is greater in this experience that in a practical

battery, more than one-hour OCV would certainly be needed to partially dissolve the SEI

and suppress its passivation in a full cell.

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 97

Figure 3. 3 Assessment of the SEI stability as function of temperature over time by mimicking its formation on inert material by cyclic voltammetry performed at 50 mV/s in 20 m LiTFSI in a 3-electrodes cell using glassy carbon as working electrode, Pt wire as counter electrode and silver wire as pseudo-reference. (a) 25 °C, (b) 35 °C, (c) 45 °C, (d) 55 °C.

Hence, prior to study the effect of resting period on the SEI degradation in battery

set-up, the impact of the electrolyte volume on the SEI dissolution was determined by

assessing discharge capacity losses during resting period. To do so, the protocol

illustrated in Figure 3. 4 was applied to check the effect of the initial volume of

electrolyte on the discharge capacity lost during resting period. Normalized discharge

capacity loss as function of electrolyte volume is shown in Figure 3. 5.

98 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

Figure 3. 4 Illustration of the protocol performed to assess the electrolyte volume influence on the SEI stability in a Mo6S8/LFP cell in 20 m LiTFSI. A pre-cycling step of 10 cycles at 1C is applied at room temperature followed by a cycling step at 0.5C with 20 h of open circuit voltage (OCV) after the second charge at room temperature.

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 99

Figure 3. 5 Influence of the electrolyte volume on the SEI stability in a Mo6S8/LFP full cell in 20 m LiTFSI. (a) Coulombic efficiency as function of cycle number. (b) Zoom on the Coulombic efficiency obtained after 20 h open circuit voltage (OCV) as function of the volume of electrolyte. (c) Normalized discharge capacity as function of the cycle number. Normalization by the 1st discharge capacity obtained at 0.5C during cycling step. (d) Zoom on the normalized discharge capacity as function of the volume of electrolyte after 20 h OCV.

While the electrolyte volume is increased from 50 µL up to 600 µL, all cells

components and parameters are kept constant, i.e., the electrode diameter and number

of glass fiber separators. Coin cells were used for electrolyte volume below 200 µL to

ensure good reproducibility of the results as the pressure applied is controlled to 0.8 T.

However, when testing greater electrolyte volumes (above 200 µL), coin cells cannot be

used since the electrolyte overflows in the crimping machine. Thus, Swagelok® design

was used despite lower performances and poorer reproducible results due to the hand-

applied pressure, as highlighted when comparing the results obtained for 500 µL volume

of electrolyte (Figure 3. 5a and c, green crosses).

100 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

First, as shown in Figure 3. 5, Coulombic efficiencies measured throughout the pre-

cycling step show a similar behavior with a steady increase from 0.91 up to 0.98 after 8

cycles, attesting the formation of the SEI regardless of the electrolyte volume. In

contrast, greater disparities are noticed for normalized discharge capacities (see Figure

3. 5c and d). Considering the limited differences observed for the values of capacity

retention obtained for coin cells and the dispersion of values in the case of Swagelok,

one can conclude that these disparities certainly originate from the difference of results

obtained between coin cell and Swagelok formats, rather than from the electrolyte

volume. This assumption is further confirmed by looking at the similar Coulombic

efficiency and normalized discharge capacity estimated after 20 h of OCV regardless of

the electrolyte volume. Moreover, considering a 10 nm thick LiF-based SEI111 forming on

a 1.27 cm diameter electrode, one can estimate the concentration of LiF expected if the

SEI dissolves as function of the electrolyte volume, as shown in Figure 3. 6a (LiF solubility

limit will be discussed in greater details in the next Chapter). Thus, even for a large

volume of electrolyte ≈ 600 µL, the solubility limit is reached and the LiF-based SEI

should not dissolve as a whole.

Figure 3. 6 (a) Estimated LiF concentration for a 10 nm LiF layer as function of the electrolyte volume (from 9.4 µL to 600 µL) considering an electrode surface of 1.27 cm² (details of calculation are given in Chapter Material and Methods). Dash line corresponds to the solubility limit determined using an ion selective electrode to fluoride, as detailed in Chapter 4. (b) Specific energy (red) and maximum number of repeated cycles (blue) (one cycle is defined by one pre-cycling step of 10 cycles performed at 1C followed by a cycling step of charge/discharge and 20 h OCV at 0.5C, as described in Figure 3. 4) before crystallization at 21 m LiTFSI as function of the electrolyte volume (from 9.4 µL to 600 µL). Energy density and specific energy correspond to calculations made with the model developed by Betz et al.221 by changing the electrolyte volume from optimized amount (9.4 µL) up to 600 µL (see Figure A3.1). Details for the calculation are given in the Appendix of this Chapter (see Table A3.6-11).

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 101

The number of repeated cycles (each cycle being composed of 10 pre-cycles at 1C

followed by 20 h OCV at 0.5C) that can be made before reaching 21 m, concentration at

which LiTFSI crystallizes at RT, are estimated as a function of the electrolyte volume. For

this calculation, two hypotheses were made to assess the amount of water consumed

during one cycle, based on the results presented in Chapter 2. First, the HER is

responsible for 70 % of the irreversible capacity during continuous cycling (see Table 2.

2 in Chapter 2). Second, 100 % of the capacity loss during OCV period is due to the HER

(see Figure 2.18 in Chapter 2). Thus, the amount of water available in the cell after each

cycle can be compared to the amount of water expected for a concentration of 21 m.

Obviously, these calculations do not consider the consumption of salt during cycling to

form the SEI. Nonetheless, as shown in Figure 3. 6b, the lower the amount of electrolyte,

the faster the 21 m LiTFSI concentration is reached and thus the faster the cell dies, as

expected. Moreover, to estimate the energy density and the specific energy as function

of the electrolyte volume, the model developed by Betz et al.221 was used. In this

protocol, values are estimated for Li-ion battery based on lab-scale measurements (in

Swagelok or coin cells) by extrapolation of the electrode materials loading and the

electrolyte volume usually employed for 18650 cells. The details for these calculations

are given in the appendix of this Chapter. Hence, increasing the electrolyte volume

obviously lowers the specific energy and the energy density due to the weight and

volume added, as shown in Figure 3. 6b and in Figure A.3.1. One can observe that using

an optimized volume of electrolyte of ≈ 10 µL (based on the electrodes porosity), 55

Wh/kg (161 Wh/L) are estimated for the battery performances. However, the cell would

reach the saturation limit after only one unique protocol of 10 charges/discharges at 1C

followed by 0.5C cycling with 20 h of resting period.

Electrochemical impedance spectroscopy (EIS) was then performed to assess the

formation of the SEI, as proposed by Suo et al.146. Indeed, by fitting the experimental

data with an equivalent circuit taking into account the electrolyte resistance, the

impedances related to the SEI, the double layer and the diffusion contribution, the

authors found an effect of cycling on the SEI formation on Mo6S8 in a 2-electrode full cell

based on Mo6S8/LMO. Unlike Suo et al.146, we used a 3-electrode Swagelok to de-

correlate contributions from both the negative and the positive electrodes. The protocol

illustrated in Figure 3. 7 was employed, using Mo6S8 as counter electrode, LFP as working

electrode and a ring of partially delithiated Li0.5FePO4 deposited on a stainless steel

gauze as reference. The electrochemical signature of the cell is given in Figure 3. 7.

Before analyzing the EIS data, one must acknowledge that faster self-discharge is

measured in this configuration when compared to the coin cell one. Observing that the

102 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

performances obtained during the pre-cycling steps is poor (see Figure 3. 7b), this

behavior certainly originates from a poorly passivating SEI formed in this configuration

which combines more separators and a lower pressure. However, despite these

experimental limitations, this protocol was used to assess the SEI stability. First, we

found that the contribution of the LFP counter electrode to the cell impedance barely

varies throughout the experiment, thus confirming that the influence of high capacity

loading of LFP versus Mo6S8 (4:1) is enough to neglect the impact of the cell state-of-

charge (SOC) on LFP impedance. Moreover, the electrochemical signature and

performances of the 3-electrode cell, despite not being optimized, are reproducible, as

shown in Figure A.3. 2 in the Appendix of Chapter 3.

Figure 3. 7 (a) Illustration of the protocol used to assess SEI stability over self-discharge cycling in a 3-electrodes cell based on Mo6S8 as counter electrode, LFP as working electrode and a ring of electrochemically delithiated Li0.5FePO4 on a stainless steel gauze as reference in 20 m LiTFSI. Cell voltage, i.e. 𝐸𝐿𝐹𝑃 − 𝐸𝑀𝑜6𝑆8, is shown. (b) Cell voltage as

function of capacity (based on Mo6S8 electrode mass).

Figure 3. 8 shows the impedance spectra obtained for Mo6S8 during cycling. First, one

can notice that Mo6S8 impedance spectra are affected by the state-of-charge at low

frequency (see Figure 3. 8b and d), which might be related to changes in diffusion in the

electrode material, among other phenomena. Moreover, the two semi-circles at middle

frequency, defined by the arrows in Figure 3. 8b, could be attributed to SEI and charge

transfer contribution146. Considering either a charged or discharged state, no changes

are visible upon cycling, as shown in Figure 3. 8b and 3. 7c for charged state and in Figure

3. 8d and e for discharged state. Besides, the resting period of 20 h does not seems to

have an influence on the response of the system, as the spectra obtained before and

after the OCV period are similar (yellow and light purple crosses in Figure 3. 8e).

Therefore, using our 3-electrode cell configuration, we conclude that EIS is not an

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 103

appropriate tool to study the SEI formation and its degradation upon cycling, unless

further optimization regarding the cell set up and reference position are made at the

negative electrode. Thus, rather than EIS, we then investigated the SEI stability by

monitoring the impact of water consumption on the electrolyte concentration.

Figure 3. 8 Nyquist plot obtained by electrochemical impedance spectroscopy (EIS) recorded for Mo6S8 upon cycling. (a) Summary of all impedance spectra. (b, c) Impedance spectra obtained in charged state. (d, e) Impedance spectra obtained in discharged state. EIS study was performed with a 20 mV amplitude signal between 1

104 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

MHz and 50 mHz of frequency range using Mo6S8 as counter electrode, LFP as working electrode and Li0.5FePO4 as reference electrode in 20 m LiTFSI.

Impact of water consumption on electrolyte crystallization

To compare and rationalize the importance of water consumption during cycling and

self-discharge to electrolyte concentration, the rate of water consumption associated

with each reaction (direct and indirect HER) was calculated taking into account the time,

the mass of active material and the hydrogen release. A volume of electrolyte of

200 µL/cell that corresponds to an initial quantity of water of 2.79 mmol (based on the

20 m LiTFSI physico-chemical properties) was used as a baseline.

Figure 3. 9 Illustration of self-discharge under open circuit voltage (OCV) happening for Mo6S8/LFP cells using 20 m LiTFSI electrolyte after being charged to 100 % state-of-charge (SOC).

Based on the cycling curve shown in Figure 3. 9, one can consider that 75 % of the

initial capacity is lost during this self-discharge protocol, bearing in mind that three

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 105

electrons are exchanged to de-insert three Li+ from Li4Mo6S8 as described by Equation

3. 1. Thus, the self-discharge mechanism can be written as follow.

𝐿𝑖4𝑀𝑜6𝑆8 → 𝐿𝑖𝑀𝑜6𝑆8 + 3 ∙ 𝐿𝑖+ + 3 ∙ 𝑒−

+

2 ∙ 𝐻2𝑂 + 2 ∙ 𝑒− → 2 ∙ 𝐻𝑂− + 1 ∙ 𝐻2

________________________________________________________

Equation 3. 1:

𝐿𝑖4𝑀𝑜6𝑆8 + ∙ 3𝐻2𝑂 → 𝐿𝑖𝑀𝑜6𝑆8 + 3 ∙ 𝐿𝑖+ + 3 ∙ 𝐻𝑂− + 3

2∙ 𝐻2

Moreover, the amount of water consumed (in mole) can be written as function of the

amount of electrode material (see Equation 3. 2 and Equation 3. 3)

Equation 3. 2:

𝑛𝐻2𝑂 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝑏𝑦 𝑠𝑒𝑙𝑓−𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒 = 3 ∙ 𝑛𝐿𝑖4𝑀𝑜6𝑆8

Equation 3. 3:

𝑛𝐻2𝑂 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝑏𝑦 𝑠𝑒𝑙𝑓−𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒 (𝑚𝑜𝑙) = 3 ∙0.75 ∙ 𝑄 (𝐶)𝑐𝑒𝑙𝑙

𝑖𝑛𝑡𝑖𝑎𝑙

𝑧 ∙ 𝐹

with 𝑄𝑐𝑒𝑙𝑙𝑖𝑛𝑡𝑖𝑎𝑙 (𝐶) the initial cell capacity calculated according to the limiting material,

i.e. Mo6S8; z the number of electron transfer to reach 75 % of delithiation, i.e. 3; F the

Faraday constant equal to 96500 C/mol.

Besides, for the direct mechanism, the rate of water consumption can be determined

by calculating the number of moles of water consumed during cycling thanks to pressure

cell experiments. As detailed in Chapter 2, the amount of mole of gas produced during

cycling is determined thanks to the Faraday law. Then, the amount of mole of water

consumed is correlated to the amount of mole of hydrogen through the Equation 3. 4.

Equation 3. 4:

𝑛𝐻2𝑂 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝑏𝑦 𝑐𝑦𝑐𝑙𝑖𝑛𝑔(𝑚𝑜𝑙) = 2 ∙ 𝑛𝑔𝑎𝑠

with ngas (mol), defined by the ideal gas law and 2, the stoichiometric coefficient

between H2 and H2O considering direct HER.

106 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

Finally, the quantity of water consumed during self-discharge or cycling can be

normalized by nwater, the initial amount of water in the electrolyte, tself-discharge/cycling, the

duration of self-discharge to lose 75 % of the capacity or the charge time

and 𝑚𝑎𝑐𝑡𝑖𝑣𝑒 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙, the mass of active material, as described in Equation 3. 5.

Equation 3. 5:

Rate of water consumed during self − discharge or cycling

= 𝑛𝐻2𝑂 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝑏𝑦 𝑠𝑒𝑙𝑓−𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒/𝑐𝑦𝑐𝑙𝑖𝑛𝑔

𝑛𝑤𝑎𝑡𝑒𝑟 ∙ 𝑚𝑎𝑐𝑡𝑖𝑣𝑒 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙 ∙ 𝑡𝑠𝑒𝑙𝑓−𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒/𝑐𝑦𝑐𝑙𝑖𝑛𝑔

Figure 3. 10 Rate of water consumed by self-discharge as function of the time spent to form the SEI.

Figure 3. 10 summarizes the rate of water consumption for cells that underwent self-

discharge protocol until 25 % SOC. One can observe that when sufficient time is spent

to form a stable SEI, i.e. at low rate and/or with multiple formation steps, the rate for

the indirect HER mechanism is constant at ≈ 0.15 %initial water amount/(hself-discharge∙gactive

material). Hence, this protocol highlights the conditions under which a passivating SEI is

formed and assesses the rate of water consumption during self-discharge. Furthermore,

this rate (indirect mechanism) is within the same order of magnitude, though smaller,

than the rate of water consumption during cycling (direct mechanism, considering SEI

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 107

formation time greater than 20 h, see Table 3. 1), suggesting that its irreversible impact

cannot be neglected when assessing the practicability of WiSE. Bearing in mind that a

20 m LiTFSI electrolyte operates close to the limit of solubility of the salt, the combine

direct and indirect water consumption rates can lead to changes in concentration and

eventually to severe drying out of the cell upon cycling and/or storage. Hence, despite

the fact that most of capacity lost during self-discharge is reversible (regarding Li+

balance, see Chapter 2), it will lead to early cell death.

Table 3. 1 Ratio of water consumed during cycling in % initial water amount /(h∙g) assessed by pressure cell measurements.

Moreover, to assess the impact of water consumption, we recorded the evolution of

water content in the electrolyte as function of cycle number and self-discharge period

experienced by a WiSE-based cell. In the literature, NMR, IR or Raman spectroscopy

results were shown to be sensitive to the salt concentration in WiSE. Though, in this

study, the use of differential scanning calorimetry (DSC) was preferred for practical

reasons. Indeed, DSC enables to limit the contamination by moisture before carrying out

the experiments (when no operando spectroscopical cells are available, as it is the case

of us) while showing great sensitivity to changes in concentration, as shown in Figure

A.3. 4.)

First, a calibration curve is determined using reference electrolytes with

concentration of 19.803 m, 20.591 m, 20.910 m, 21.992 m and 22.9983 m. Using the

first crystallization peak of the reference electrolytes (described by the arrow in Figure

A.3. 4b), the following fit 𝑇𝑐𝑟𝑦𝑠𝑡𝑎𝑙𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 𝑝𝑒𝑎𝑘 = 4.68 ∙ 𝐶𝐿𝑖𝑇𝐹𝑆𝐼𝑚𝑜𝑙 𝑘𝑔⁄

− 91.5 (see Figure 3.

11c) is obtained.

Then, DSC experiments were performed on separators collected from aged cells, as

shown in Figure 3. 11c and d. Looking at the exothermic peak corresponding to the

hydrated LiTFSI phase (H2O∙LiTFSI) crystallization, cells can be split into two groups: 1)

those that underwent continuous cycling (yellow and orange curve) at lower

Cycling (results based on pressure cells) 0.1C 0.15C

Mass of active material [mg] 9.63 7.88

Amount of hydrogen produced/cycle [µmol] 1.43 0.978

Duration of cycling (related to SEI formation) [h]

40 26.7

Amount of water consumed during cycling/cycle [µmol]

2.86 1.96

Rate of water consumption [%/h/g] 0.27 0.33

108 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

temperature and 2) the cells that underwent a 20 h OCV step after each charge (pink

and purple line). Based on these results, one can hypothesize that, upon repeated

resting period, the electrolyte salt concentration increases, originating from a non-

negligible amount of water consumed, thus confirming our previous calculation of the

water consumption rate. Eventually, a greater salt concentration is deduced for cells

cycled during 15 cycles suggesting a greater water consumption than in the case of cells

cycled during 100 cycles (see Figure 3. 11c). This observation could originate from partial

water recombination during cycling, however, reproducibility must be verified before

drawing any definitive conclusion. Calibration curves and additional measurements on

cells aged are currently under investigation. To complement these results and accurately

reflect the cell drying off, further measurements can be done by fixing the overall time

of experiment while changing the cycling or self-discharge cycle number.

Finally, one has to acknowledge that assessing water consumption in coin cell devices

still remain far from 18650 or pouch cells in which the design will drastically modify the

electrolyte/electrode ratio and so, the influence of the parasitic reactions.

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 109

Figure 3. 11 Differential scanning calorimetry (DSC) experiments performed at 2 °C/min between 35 °C and -60 °C upon cooling and back to 60 °C on heating. (a) Cells aged for 15 and 100 cycles with continuous cycling protocol or with a 20 h OCV resting period at the end of each charge (b) zoom on the crystallization peak of aged cells. (c) Crystallization peak measured as function of electrolyte concentration obtained from the calibration data (black cross) and linear fit (black line) enabling the calculation of the aged cells concentration (red cross).

Activation energy of direct and indirect HER in WiSE

As shown by self-discharge measurements, the SEI formed upon cycling in WiSE-

based electrolyte does not prevent water to access the interface where it is reduced.

Our results suggest a much higher HER onset potential than the one previously

determined (1.9 V vs Li+/Li)111 by linear scan voltammetry on stainless steel current

collectors. Knowing that the potential for de-intercalation of Li4Mo6S8 is ≈ 2.7 V vs Li+/Li,

the reversible potential for water reduction must be greater than this value to explain

self-discharge observed in this work (see Figure 3. 12a). This conclusion is in line with

the results published by Kühnel et al.153 assessing the ESW as function of the current

density threshold with an onset potential for HER at 2.48 V vs Li+/Li (ithreshold = 2 µA/cm²)

110 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

on stainless steel. Besides, HER potential also depends on the electrode material153. Naik

et al.222 reported a HER onset potential on Mo6S8 to be around -0.2 V vs NHE, i.e. ≈ 2.8 V

vs Li+/Li in diluted media, barely dependent on the pH of the electrolyte.

Furthermore, by normalizing the time needed to fully de-insert Li4Mo6S8 by constant-

current de-insertion or by de-insertion through self-discharge, two de-insertion distinct

rates can be seen (see Figure 3. 12b). Indeed, during self-discharge, three Li+ are de-

inserted in ≈ 1/3 of the overall time during the 2nd plateau while one Li+ is de-inserted in

≈ 2/3 of the time during the first plateau, indicating that the de-intercalation of Li+ is

faster during the second plateau when driven by self-discharge. To explain this

observation, one must first keep in mind that the driving force for de-intercalation is

different between the two plateaus. Hence, the driving force defined by ΔE = EHER – Ede-

intercalation is greater for the second plateau at 2.3 V than the first plateau at 2.7 V vs Li+/Li

(see Figure 3. 12a), which eventually lowers the activation barrier for the indirect HER

associated with the second plateau.

Figure 3. 12 (a) Redox potentials for Mo6S8 and the HER, extracted from the literature111,222. (b) Voltage as function of normalized time for a Mo6S8/LFP cell in 20 m LiTFSI undergoing self-discharge (dotted line) or constant-current continuous cycling (full line). Normalized time is defined by the ratio between evolution of time during discharge and the end of discharge time.

In order to clarify these observations, activation energies for self-discharge

associated with the first and the second intercalation plateau of Mo6S8 in WiSE were

determined using a Mo6S8/LFP cell based on 20 m LiTFSI electrolyte. To perform this

study, (i) the LFP electrode is in large excess compared to the Mo6S8 one, (ii) oxygen is

purged from the electrolyte before assembly and (iii) the contribution of LFP to the self-

discharge is considered negligible, as demonstrated in Chapter 2 where water oxidation

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 111

potential was shown to occur above 4 V vs Li+/Li. In addition, we considered that the

reaction between charged (delithiated) FePO4 and hydrogen produced at the negative

electrode is negligible, as it is controlled by the diffusion of hydrogen across the cell.

To determine these activation energies, the protocol described in Figure 3. 13a and c

was used. First, a pre-cycling step of 10 cycles performed at 1C, 25 °C, was applied to

form a sufficiently stable SEI (as concluded from Figure 3. 10). The cell was then cycled

at 0.5C at different temperatures (15 °C, 25 °C, 35 °C, 45 °C, 55 °C) using a

charge/discharge/charge protocol followed by 20 h OCV. Capacity retention was

calculated by normalizing the discharge capacity obtained after 20 h of resting period to

the one obtained during full cycling. For the 1st plateau of Mo6S8 (2.7 V vs Li+/Li), a cut

off of 1.2 V was set during the cycling step, while a cut off of 1.5 V was used for the 2nd

plateau. Then, following the Arrhenius equation, the normalized loss of discharge

capacity upon self-discharge is plotted as function of the inverse of temperature (Figure

3. 13.b and d), allowing the extraction of the activation energies from Equation 3.6.

Equation 3. 6:

𝑘 = 𝐴 ∙ 𝑒−𝐸𝑎𝑅𝑇

with k, the reaction rate (the unit depends on the reaction considered, no unit for

self-discharge, h-1 for pressure cell and A for CC-CV 2-electrodes coin cell); A the pre-

exponential factor (the unit is defined by the unit of the reaction rate); Ea, the activation

energy (J/mol); R, the universal gas constant (8.314 J/mol/K) and T, the temperature (K).

Almost identical activation energies are extracted following this protocol (19 kJ/mol

for the 2nd plateau and 25 kJ/mol for the 1st plateau), regardless of the potential of the

intercalation plateau (Figure 3. 13), suggesting that self-discharge mechanism is similar

between these two insertion plateaus, the difference of reaction rates being attributed

to difference in driving force (ΔE = EHER – Ede-intercalation).

112 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

Figure 3. 13 Determination of the activation energy of the indirect HER taking place during self-discharge. (a) Potential as function of time for self-discharge experiment performed on the first plateau of Mo6S8 (2.7 V vs Li+/Li). Ten cycles are performed at 1C, 25 °C, followed by cycling at 0.5C at different temperature (15 °C, 25 °C, 35 °C and 45 °C). (b) Arrhenius plot of the normalized discharge capacity loss by indirect HER during self-discharge as function of temperature. Unfortunately, no value was determined at 55 °C for the first plateau, since changes in the electrochemical signature are observed and are currently investigated. (c) Potential as function of time for self-discharge experiment performed on the second plateau of Mo6S8 (2.3 V vs Li+/Li). Ten cycles are performed at 1C, 25 °C, followed by cycling at 0.5C at different temperature (15 °C, 25 °C, 35 °C, 45 °C and 55 °C) (d) Arrhenius plot of the normalized discharge capacity loss by indirect HER during self-discharge as function of temperature. (b, d) Data points represent the mean of three cells, the exception being only one cell at 15 °C during 1st plateau experiment. All cells are based on Mo6S8/LFP cell with 20 m LiTFSI electrolyte.

Moreover, self-discharge reaction (see Equation 3. 7) is a combination of two redox

reactions (see Equation 3. 8 and Equation 3. 9) that comprised multiple steps, as more

than one electron is transferred. Besides, ions exchange (HO- and Li+) can lead to mass

transport limitation in the bulk of the electrolyte, across the SEI and/or in the electrode

material. Therefore, comparison between the activation energy obtained for self-

discharge should be confronted with activation energies for these phenomena.

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 113

Equation 3. 7:

𝐿𝑖4𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐻2𝑂 → 𝐿𝑖4−𝑥𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐿𝑖+ + 𝑥 ∙ 𝐻𝑂− + 𝑥

2∙ 𝐻2

Equation 3. 8:

𝐿𝑖4𝑀𝑜6𝑆8 → 𝐿𝑖4−𝑥𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐿𝑖+ + 𝑥 ∙ 𝑒− 𝑤𝑖𝑡ℎ 𝑥 ≤ 4

Equation 3. 9:

𝑥 ∙ 𝐻2𝑂 + 𝑥 ∙ 𝑒− → 𝑥 ∙ 𝐻𝑂− +𝑥

2∙ 𝐻2 𝑤𝑖𝑡ℎ 𝑥 ≤ 2

Considering the redox reaction associated with lithium de-intercalation (Equation 3.

8), mass transport is usually not considered as rate determining step (for moderate C-

rate such as those used in this study) since Li+ diffusion in Mo6S8 material was reported

to be fast enough during intercalation/de-intercalation process with an activation

energy comprised between 10 to 17 kJ/mol in Mo6S8223,224. Furthermore, diffusion in the

SEI is often considered as fast compared to charge transfer225. Eventually, bulk diffusion

in the electrolyte is generally not limiting the intercalation mechanism at moderate C-

rate due to a rather high conductivity of the electrolyte (≈10 mS/cm at 25 °C for 21 m

LiTFSI). Therefore, the rate-determining step for intercalation reaction in LIB is generally

considered to be the de-solvation step during charge transfer226, which activation

energy largely varies with the solvent. Indeed, it was reported to be much faster in

diluted aqueous-based electrolyte75,227 than in organic based one225,228. Unfortunately,

the impact of superconcentration and thus of solvation sheath on the activation energy

of the de-solvation process is barely studied, with the exception of Hu et al. 229 who

recently reported an activation energy for solvation of 20 kJ/mol in 20 m LiTFSI.

Nonetheless, such value is highly debatable due to questionable experimental methods

that include measurements at 0 °C, temperature at which the electrolyte is crystallized.

Altogether, one can hypothesize that the solvation/desolvation process should limit,

alike in organic based LIB, the intercalation reaction in WiSE-based LIB systems.

However, the determination of the rate-determining step in the overall self-discharge

reaction is not straightforward since both half-redox reactions (see Equation 3. 8 and

Equation 3. 9) may interplay.

Attempts were thus made to determine the activation energy of the HER in a battery

device. To do so, Mo6S8/LFP pressure cells were assembled using 20 m LiTFSI electrolyte.

The SEI was first formed by performing 10 cycles at 1C, 25 °C. The activation energy for

the HER was then extracted by measuring the pressure increase at different

temperatures (25 °C, 35 °C, 45 °C and 55 °C) during the four subsequent continuous

114 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

cycles performed at low C-rate (0.15C), as illustrated in Figure 3. 14a. To determine the

activation energy of the direct/electrochemical HER, the rate of hydrogen production

was determined considering that gas is generated only during charge (see Figure A.3. 5

of the Appendix of this chapter). Such experiments are inspired from the work done by

Ayeb et al.230 who determined by gases measurements the HER activation energy during

overdischarge conditions in Ni-MH battery to be in the range from 50 to 70 kJ/mol.

Figure 3. 14b shows the Arrhenius plot obtained for the direct HER on the surface of

Mo6S8, from which an activation energy of 72 kJ/mol is determined. Unfortunately, no

values for the HER activation energy on Mo6S8 was reported for comparison, to the best

of our knowledge. However, Mo6S8 is reported to be a very promising HER catalyst in

acid222,231,232 with activity reaching the one of Pt -which is known to have an activation

energy of about 10 kJ/mol in acid -, albeit the HER is known to be much slower at pH

higher than 2, corresponding to our experimental conditions233. Thus, one can

hypothesize that the activation energy obtained in this study is rather high and

surprising. This can be explained by the use of pressure cell-type devices, in which the

activation energy extracted from pressure measurements are obtained in a regime in

which with mass transport limitations occur. Moreover, experimentally speaking, due to

the low C-rate used (0.15C), a plateau corresponding to pure direct HER is visible above

1.4 V (and before the 1.5 V cut off) at 45 °C and 55 °C, as shown in the appendix of this

chapter. Finally, and more importantly, as both HER and intercalation reaction take place

simultaneously, the total current is not entirely directed toward the HER.

Figure 3. 14 (a) Illustration of the protocol used to assess the activation energy of direct HER on Mo6S8 in 20 m LiTFSI by pressure cell tests. Potential and pressure as function of time. (b) Arrhenius plot showing the pressure increase as function of temperature.

To circumvent all these uncertainties related to the experimental conditions, we

completed by measuring the activation energy for direct HER in practical conditions

applying a constant current - constant voltage (CCCV) protocol to reach the HER

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 115

potential after full charging of the cell, as illustrated in Figure 3. 15a. In details, 2-

electrodes coin cells were assembled and pre-cycled at 25 °C during 10 cycles at 1C to

form the SEI. The cell was then charged to 100 % SOC with a voltage cut off fixed to

1.4 V, to avoid any pure HER plateau starting below 1.5 V above 45 °C. Once 100 % SOC

is reached, the cell voltage is set to 1.5 V for 20 h. The current density was then

measured as function of temperature (25 °C, 35 °C and 45 °C and 55 °C). The

corresponding Arrhenius plot is shown in Figure 3. 15b and the activation energy for

pure HER on Mo6S8 in 20 m LiTFSI in battery set up is found to be 25.5 kJ/mol.

Figure 3. 15 (a) Protocol used to assess the activation energy of direct HER for Mo6S8 in 20 m LiTFSI in coin cells. Potential as function of time with pre-cyling step performed at 1C at 25 °C in blue, and cycling step based on full charge at 0.5C up to 1.4 V followed by the application of 1.5 V during 20 h as function of temperature (25 °C, 35 °C, 45 °C and 55 °C) in red as well as current response to a 1.5 V voltage applied (in yellow). (b) Arrhenius plot for the current measured at 1.5 V and 100 % state-of-charge as function of temperature (each point is the mean of two cells).

Although the value measured in this study has some uncertainty and the self-

discharge experiments measure a global phenomenon, it represents the actual

activation energies in practical conditions. Comparing the activation energy found for

self-discharge occurring during both plateau with that measured during the CCCV

measurements, values are found very similar (≈ 20 kJ/mol). Therefore, we can conclude

that the kinetics for the self-discharge mechanism is presumably governed by the HER

rate. For the sake of comparison, activation energies determined for self-discharge for

Lead-acid battery and Ni-MH battery (considering self-discharge driven by water

reacting with the positive electrode, see Equation 3. 10: and Equation 3. 11:) or organic-

electrolyte-based LIB, are listed in Table 3. 2 and compared to the experimental value

obtained in this study, regardless of the operating conditions used.

116 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

Table 3. 2 Summary of activation energy corresponding to the overall self-discharge mechanism. Considering Ni-MH and Lead-acid, the reaction taken into account is the reaction with water of the positive electrode at 100 % SOC, as described in Equation 3. 10 and Equation 3. 11. Considering LIB, the positive electrode or the negative electrode are considered as limiting electrode.

Considering lead-acid and Ni-MH devices, the positive electrode is the capacity-

limiting one, unlike in our system where the limiting electrode is the negative. This

technological choice is made such that oxygen is evolved preferentially instead of

hydrogen upon overcharges or self-discharges (see Equation 3. 10 and Equation 3. 11).

Having the OER as the rate-determining step238,239 reduces the self-discharge rate

compared to an HER-driven one since OER kinetics is more sluggish. Hence, the

activation energies measured for WiSE (20 kJ/mol) is much lower than those reported

for Ni-MH and Lead-acid (67 and 40-70 kJ/mol, respectively).

Equation 3. 10:

𝑃𝑏𝑂2 + 2𝐻+ + 𝑆𝑂42− = 𝑃𝑏𝑆𝑂4 + 𝐻2𝑂 +

1

2𝑂2

.

Equation 3. 11:

𝑁𝑖𝑂𝑂𝐻 + 1

2𝐻2𝑂 = 𝑁𝑖(𝑂𝐻)2 +

1

4𝑂2

For the LIB technology, two ranges of values were reported for the activation energy

related to self-discharge: low values of ≈ 20 kJ/mol and greater values up to 80 kJ/mol.

Activation energy of 20 kJ/mol is observed at the negative electrode (synthetic flake

graphite)234 and linked to self-discharge governed by Li+ de-intercalation and SEI growth,

both associated to electrolyte reduction (1.2 M LiPF6 in EC:DEC (1:3 vol%)). The greater

values are related to Li+ intercalation occurring at the positive electrode (LCO) and

associated with electrolyte oxidation. These comparisons have a direct impact on

practicality of the WiSE technology, as it traduces that the shelf-life of systems for which

the self-discharge is governed by reactions at the negative electrode will be greatly

limited. To summarize, for WiSE-based LIB, the self-discharge mechanism seems to be

controlled by fast kinetics reaction occurring at the negative. However, the strategies

employed in other aqueous-based batteries in which the positive produces oxygen and

Technology This work Organic LIB Ni-MH Lead-acid

Activation energy (kJ/mol)

≈ 20 20234 to 80 235,236 67 ref 237 40-70 ref 238

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 117

prevents hydrogen evolution, i.e. Lead-acid and Ni-MH, cannot be transposed as

discussed in the following section.

Discussion and conclusion of the chapter

In this chapter, we monitored the instability of the native SEI formed on negative

electrode when cycled in aqueous superconcentrated electrolyte (20 m LiTFSI). First, as

shown on a model-electrode (glassy carbon) and further confirmed by self-discharge

experiments, parasitic reactions are not prevented during resting period by the

protective passivating layer initially formed. DSC experiments also attest the irreversible

consumption of water during both cycling and self-discharge experiments. In addition,

based on self-discharge and continuous cycling experiments, one can determine that the

rate of water consumption is in the same order of magnitude (in terms of %initial water

amount consumed/(h∙gAM)) during self-discharge at 2.3 V vs Li+/Li than during constant current

continuous cycling, though smaller. Besides, as shown by normalizing the time of Li+ de-

insertion either during cycling or by self-discharge, reaction rates are found different. To

understand this observation, we designed electrochemical protocols to assess the

activation energies associated with self-discharge and direct HER. Experimental results

confirm a similar activation energy for self-discharge on both Mo6S8 plateau, the

difference in reaction rate being explained by differences in driving force. Despite

difficulties in accurately estimating the activation energy for direct HER, our results

suggest that the rate for self-discharge is governed by the generation of hydrogen, and

not by delithiation of Mo6S8, explaining the difference in de-intercalation rate.

Altogether, we highlight the dramatic impact of self-discharge that is driven by parasitic

reactions occurring at the negative electrode on the practicality of this technology.

To conclude on the viability of WiSE-based LIB, the figure of merits for WiSE-based

aqueous Li-ion system is compared to those for classical organic Li-ion

(NMC111 / LP30 / graphite) as well as other aqueous systems (see Figure 3. 16). As

evidenced in our study, WiSE-based aqueous batteries can only safely operate within a

≈ 2 V operating window to avoid parasitic side reactions, unlike organic LIB. To be able

to compare our WiSE or WiBS-based battery to classical Li-ion or commercial aqueous

systems, we estimated both the energy density and the specific energy for WiSE and

WiBS following the protocol proposed by Betz et al.221. In conclusion, both the specific

energy and the energy density at a cell level are twice smaller than for aprotic Li-ion

batteries and eventually similar to the ones achieved by Ni-MH batteries, while being

above those for Ni-Cd or Lead-acid batteries. However, the energy efficiency is

118 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

similar/close to the one obtained for Li-ion (> 90 %), unlike Lead-acid, Ni-Cd, and Ni-MH

batteries that show energy efficiency closer to 80-85 %.

Figure 3. 16 Benchmarking WiSE-based and WiBS-based aqueous batteries against other aqueous systems, namely Lead-acid, Ni-Cd and Ni metal hydrides (Ni-MH) batteries as well as against aprotic Li-ion batteries. The spider-chart at the center compares these systems in terms of six parameters defining the overall performances of these systems. On top is compared the specific energy for these systems, on the left the self-discharge for these systems, on the right the energy density, on the bottom left is represented the specific energy as a function of specific power, on the bottom right is reported the energy efficiency for these systems while on the bottom the operating temperature window for the different technologies. All references are given in the Appendix of the Chapter.

Evidently, cycling batteries within the practical thermodynamically stable potential

window of WiSE (≈ 2 V) drastically limits the energy density, which cannot reach the one

achieved by LIB. The obvious way to increase the energy density would be to further

extend the operating window, by finding chemical-engineering means to handle the gas

generated during cycling, alike in other aqueous batteries (Lead-acid, Ni-Cd or Ni-MH).

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 119

To do so, two cell designs are generally considered for the aqueous technologies:

flooded batteries and sealed ones (that can be vented or not). For flooded batteries,

gases are evacuated away from the electrodes prior to be recombined (H2 + ½ O2 = H2O).

Furthermore, electrolyte (or water) can be added to compensate water consumption.

Regarding vented-sealed batteries, thanks to a pressure valve, the pressure increase

resulting from the HER and/or OER can be regulated by outflowing the gases from the

cell without allowing oxygen to come in. This strategy reduces the pressure build up

during the battery operation. For example, for Lead-acid batteries, the so-called Valve

Regulated Lead Acid (VRLA) sealed batteries were developed. Moreover, a catalyst such

as Pd can be added in the form of a battery plug to catalyze the gas recombination

reaction (H2 + ½ O2 = H2O) and minimize the electrolyte loss and thus the drying-out of

the cell. Hence, vented-WiSE batteries could be envisioned, providing that the salt

crystallization issue discussed by Kühnel et al.153 can be solved for superconcentrated

electrolytes, as upon continuous consumption of water the battery lifetime will rapidly

reduce.

An alternative approach to circumvent the electrolyte drying-out can be the design

of sealed WiSE-based batteries using the oxygen recombination strategy. Indeed,

another mean to control water decomposition is by adjusting the capacity balance to

promote OER compared to HER. For a capacity ratio of negative to positive (N/P) equal

to one, both the OER and the HER take place on the positive electrode and the negative

one, respectively, upon overcharge. In order to limit the hydrogen production, the

positive electrode is set to be the limiting one (N/P > 1), reaching full charge before the

negative faces HER. Hence, only O2 is produced and diffuses through the electrolyte to

recombine into water by reacting with the negative electrode (as described in Figure 3.

17), thus forming the “oxygen cycle”7 that prevents the cell drying off. Altogether, one

should keep in mind that such solution can only be efficient if the O2 production rate is

not faster than the recombination one. In definitive, if no care about the dimensionality

of the electrodes is taken, aqueous-based batteries face an unbalanced generation of

gases and poor recombination efficiency. Eventually, in order to (i) reduce electrolyte

evaporation, (ii) to enhance safety by limiting electrolyte creepage and (iii) to promote

oxygen diffusion to the negative electrode, electrolyte properties can separately be

improved. Porous gel-type electrolytes were implemented to promote the diffusion of

O2 to the negative electrode as gases diffuse faster in gel-type electrolyte than in liquid

aqueous electrolytes. Indeed, water consumption taking place in the electrolyte induces

the formation of cracks thus creating a network for O2 diffusion. Another alternative was

120 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

the use of electrolyte absorptive glass fiber separators, to avoid having excess of

electrolyte in the cells.

Figure 3. 17 Some parasitic reactions taking place in lead-acid, Ni-Cd and Ni-MH batteries. N/P ratio correspond to the capacity balance between the positive and the negative electrode.

However, altogether the development of these different concepts would (i) need

electrolyte refill to avoid drying-out of the cell or (ii) require the use of catalysts to

recombine water from O2(g) and H2(g) and/or (iii) of a gelified electrolyte to improve gas

diffusion. Therefore, these constraints would impose drastic limitations regarding both

the operating voltage as well as the charging rate for WiSE-based aqueous batteries.

Indeed, both increasing the operating voltage and increasing the charging rate would

lead to increased kinetics for gas generation. For instance, as observed for Ni-Cd

batteries, water recombination catalysts can only handle a certain amount of gas, which

limits the charging to 0.1C. Moreover, the water recombination in these aqueous

devices requires a sufficient amount of oxygen to allow recombination. Indeed, these

systems rely on the paradox that a too good positive electrode for which no oxygen is

released would not allow for the water recombination at the negative electrode.

However, as evidenced in our study, while hydrogen is produced in parallel with Li+

intercalation at the negative electrode during charge, almost no oxygen is released at

the positive electrode at room temperature. This unbalanced generation of gases

between the positive and the negative electrode towards the generation of hydrogen,

which is not encountered for other aqueous devices, drastically limits the possibility of

water recombination in WiSE-based aqueous batteries, eventually leading to severe

drying-out issues for the battery.

Chapter 3 – Instability of native SEI leads to the drying out of the LIB 121

Furthermore, another major drawback of WiSE-based batteries is the narrow

operating temperature range. Indeed, crystallization of the electrolyte may occur at

room temperature. Even though several anions have been considered to downshift the

crystallization point180,181,183,187, as discussed in chapter 1, the question of the cost,

scalability and the competitiveness of the superconcentrated aqueous electrolyte will

be raised. With this in mind, reducing further the amount of water by increasing the

amount of lithium salt in the electrolyte does not appear as a viable solution for

applications competing with Li-ion batteries since the SEI instability is not suppressed by

the use of bisalt superconcentrated electrolytes.

Finally, for grid applications, the capacity fading of WiSE-based batteries must be

limited over time. However, as shown in our work, the second intercalation plateau of

Mo6S8 corresponding to 75 % of the total capacity of the material is completely loss after

a resting period of 300 to 800 hours, depending on the C-rate employed during cycling.

Thus, capacity fading in the order of 30 to 75% per month is deduced for this specific

configuration, which must be compared with capacity fading of 2-10 % measured for

aprotic Li-ion batteries, 1-15 % for Lead-acid batteries, 15-30 % for Ni-MH and 10-20 %

for Ni-Cd, bearing in mind than some can be recovered for commercial Ni-Cd and Ni-MH

aqueous systems as well as for WiSE-based system. Thus, without further improvement

to the current technology and the finding of optimized pre-cycling conditions, WiSE-

based aqueous batteries cannot currently be regarded as a viable option for grid

electricity storage.

As a summary, the SEI formed in these superconcentrated electrolytes is not

protective enough to prevent the electrolyte degradation during cycling and resting

period, more specifically to avoid water reduction and hydrogen generation. Based on

these experimental results, we provide the figure or merit for WiSE-based Li-ion battery

that we compared to that of classical Li-ion battery and commercial aqueous system

such as Lead-acid, Ni-Cd or Ni-MH. Doing so, it clearly appears that while WiSE-based

batteries share the energy efficiency of aprotic Li-ion batteries, owing to similar

intercalation reactions, it only offers the energy density and the specific energy of Ni-

MH batteries. However, WiSE-based batteries show poorer temperature stability than

either systems. Furthermore, the continuous electrolyte consumption occurring both on

charge and upon self-discharge for WiSE-based batteries may lead to the drying-out of

the cell. Our analysis further revealed that owing to the lack of oxygen generation upon

charge, the implementation of a gas recombination cycle in WiSE-based cells as used for

other aqueous battery applications might be complex. Therefore, superconcentrated

122 Chapter 3 – Instability of native SEI leads to the drying out of the LIB

aqueous electrolyte are currently not able to compete with commercialized aqueous

systems for grid storage application until means to prevent the HER at the negative

electrode can be found and benchmarked in practical conditions. This calls for the design

of stable SEI as reported. Toward this goal, Chapter 4 will focus on the stability of

artificial SEI based either on inorganic coatings while the perspectives of this thesis will

tackle polymeric ones.

CHAPTER 4 –MIMICKING INORGANIC-BASED SEI

WITH LIF-COATING. UNDERSTANDING OF

INORGANIC SEI LIMITATIONS IN WATER-IN-SALT ELECTROLYTE2.

2 This chapter is based on the article that I co-authored: Droguet, L.; Hobold, G. M.; Lagadec, M. F.; Guo, R.; Lethien, C.; Hallot, M.; Fontaine, O.; Tarascon, J.-M.; Gallant, B. M.; Grimaud, A. Can an Inorganic Coating Serve as Stable SEI for Aqueous Superconcentrated Electrolytes? ACS Energy Lett. 2021, 6 (7), 2575–2583. https://doi.org/10.1021/acsenergylett.1c01097.

126 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Introduction

Developing stable SEIs in aqueous media has been at the forefront of research for

aqueous Li-ion batteries (LIB) and other aqueous technologies such as electrochemical

capacitors (based on carbon or pseudocapacitive electrodes) or Na-ion

batteries58,111,117,164. As described in Chapter 1, one major advance was the introduction

of superconcentrated aqueous electrolytes in which salt molecules exceed water

molecules both in volume and in mass. Indeed, in these so-called WiSE or WiBS58,122,

almost all water molecules contribute to the solvation sphere of Li+, allowing for a strong

ion-pair interaction between solvated Li+ cations and organic anions such as TFSI- or its

derivatives. However, as shown in Chapter 2 and 3 of this manuscript, the atypical

solvation structure does not suppress the reduction of water at the negative electrode.

Nonetheless, the reactivity of the -CF3 terminal groups of TFSI- anions present at the

interface is unlocked, which triggers the formation of LiF and passivate, at least partially,

the negative electrode134,146–148. Therefore, contrary to organic LIB electrolytes, where

the decomposition of both the solvent molecules and salt anions results in a native SEI

layer containing both inorganic (LiF, Li2CO3, Li2O, etc.) and organic/oligomeric

species240,241, for superconcentrated aqueous electrolytes, the passivation

predominantly relies on LiF.

The SEI thickness and density generally vary with the formation step (cycling rate,

duration, temperature etc.). Moreover, part of the SEI can be dissolved. This raises the

question of whether the water reduction observed in the previous chapters upon rest

originates from the morphology and porosity of the deposited LiF layer or from its

intrinsic solubility and/or instability in aqueous superconcentrated electrolyte. To

decide between these two options, in this Chapter, we first assessed the LiF solubility

limit in aqueous electrolytes as a function of LiTFSI salt concentration. Then, we studied

the stability of a conformal LiF layer deposited onto metallic Li anode as a proxy to mimic

the formation of a native SEI. First, by using environmental scanning electron

microscopy (E-SEM), we tested LiF protection against moisture. Second, we assessed its

stability against two aqueous superconcentrated electrolytes (20 m LiTFSI and 20 m

LiTFSI : 8 m LiBETI) using gas chromatography and compared it to insoluble Al2O3 coating

prepared onto metallic Li. Eventually, by using presoaking step in organic

solvent/electrolyte, we analyzed the importance of salt/solvent-derived species to

ensure self-passivation and maintain stable the SEI. The outline of this chapter is

illustrated in Figure 4. 1.

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 127

Throughout this Chapter, because of practical considerations in terms of energy

density gain, we focused our attention on metallic Li anode rather than on the classically

used Mo6S8 anodes, bearing in mind that the kinetics for electrolyte reduction will be

greatly enhanced together with accompanied gas generation. Hence, it provides an

accelerated approach to pinpoint the weaknesses of any SEI layer composed of LiF which

will eventually appear during cycling of anode materials.

Figure 4. 1 Illustration of the outline of Chapter 4 based on the results discussed in chapter 2 and 3.

LiF solubility limit in aqueous superconcentrated electrolyte

The solubility limit of lithium fluoride is well-known in pure solvents (note that

solubility limit measurements are performed at thermodynamic equilibrium). It is

notably greater in water than in most carbonate solvents, with the exception of EC (see

Table 4. 1).

128 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Table 4. 1 LiF solubility limit found in literature in various solvents: water, ethylene carbonate (EC), dimethyl carbonate (DMC), propylene carbonate (PC). One can note that the difference in solubility limit of two orders of magnitude in DMC was explained by differences in the protocol, especially regarding the filtration technique as detailed in Chapter Materials and Methods.

Solvent LiF solubility at 25 °C (g/L) LiF solubility at 40 °C (g/L)

H2O 1.11 242 1.20 242

DMC 0.57 242 4 ⋅ 10-3 243

PC 0.14 242 0.21 242

EC 5.52 242

Nonetheless, very little is known regarding the solubility limit of LiF as a function of

salt concentration. The common-ion effect arising from the dissolved Li-salt is believed

to lower the solubility of LiF in superconcentrated aqueous electrolytes. In order to

probe this effect, the solubility limit of LiF was directly measured in aqueous electrolytes

as a function of the LiTFSI salt concentration using a fluoride ion selective electrode (ISE)

and following the protocol developed by Strmcnik et al.244 illustrated in Figure 4. 2a

(details are given in Chapter Materials and Methods). As shown in Figure 4. 2b, the LiF

solubility limit was observed to decrease from pure water (0.93 g/L) down to 1.9 ∙ 10-3

and 1.5 ∙ 10-3 g/L for 20 m LiTFSI and 20 m LiTFSI : 8 m LiBETI aqueous electrolytes at 23

°C, respectively.

Figure 4. 2 (a) Schematic explanation of the protocol used to measure LiF solubility, further details are provided in the chapter Materials and Methods. (b) Solubility limit of LiF in saturated aqueous solutions measured in pure water and at different LiTFSI concentration ranging from 1m to 20m (WiSE) as well as for 20 m LiTFSI : 8 m LiBETI (WiBS). The dashed line is a guide to the eyes.

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 129

Considering the reaction described in Equation 4. 1, the constant of dissolution of LiF

–Ks- in an aqueous solution can be expressed following Equation 4. 2.

Equation 4. 1:

𝐿𝑖𝐹(𝑠) = 𝐿𝑖+ (𝑎𝑞) + 𝐹− (𝑎𝑞)

Equation 4. 2:

𝐾𝑠 = 𝑎𝐿𝑖+𝑎𝐹− = 𝛾𝐿𝑖+ [𝐿𝑖+] ∙ 𝛾𝐹−[𝐹

−] = 𝛾𝐿𝑖+𝛾𝐹− ∙ [𝐿𝑖+][𝐹−]

Due to LiF stoichiometry, the product of the activity coefficients of ions in solution –

𝛾𝐿𝑖+𝛾𝐹− can be defined as the square of the mean ionic activity coefficient of the salt245

i.e. �̅�𝐿𝑖+𝐹−² as expressed in Equation 4. 3. Therefore, Equation 4. 2 becomes Equation 4.

4.

Equation 4. 3:

𝛾𝐿𝑖+𝛾𝐹− = �̅�𝐿𝑖+𝐹−²

Equation 4. 4:

𝐾𝑠 = �̅�𝐿𝑖+𝐹−² ∙ [𝐿𝑖+][𝐹−]

Besides, the fluoride and lithium ions concentration can be expressed according to

Equation 4. 5.

Equation 4. 5

[𝐿𝑖+] = [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 + [𝐿𝑖+]𝑙𝑖𝑚 and [𝐹−] = [𝐹−]lim

with [𝐿𝑖+]𝑙𝑖𝑚 = [𝐹−]𝑙𝑖𝑚

where [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 (mol/L) is the initial concentration of Li+ cation due to the

LiTFSI concentration in solution (i.e. the molarities corresponding to the following

molalities: 1 m, 3 m, 5 m, 7 m, 10 m, 15 m, 20 m), [F-]lim (mol/L) is the solubility limit of

fluoride measured by the ISE electrode and [Li+]lim (mol/L) is the concentration of Li+

cation added during the LiF dissolution in the electrolyte. Considering the high initial

[Li+]electrolyte in the bulk electrolyte (> 1 m i.e. 0.87 M) and the measured fluoride solubility

limit (< 1 g / L i.e. 0.05 mol / L ), [Li+] can be simplified as follow:

Equation 4. 6:

[𝐿𝑖+] = [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒

Therefore, Equation 4. 4 is simplified in Equation 4. 7.

130 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Equation 4. 7:

[𝐹−]𝑙𝑖𝑚 =𝐾𝑠

�̅�𝐿𝑖+𝐹−² ∙ [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒

with Ks the constant of solubility, �̅�𝐿𝑖+𝐹−the mean activity coefficient of the LiF salt

and [Li+]electrolyte (mol/L) the initial concentration of lithium in the electrolyte.

As theoretically described by McEldrew et al.246, the logarithm of the mean activity

coefficient log(�̅�𝐿𝑖+𝐹−) increases with the molality. Therefore, even without knowing the

effect of the molality and thus of the solvation structure on Ks, the decrease of solubility

limit observed in Figure 4. 2b is consistent with the common ion effect and the increase

of activity coefficient, which are both related to the increase of molality. Moreover,

taking into consideration these theoretical results by McEldrew et al.246, one can rescale

the logarithm of the activity coefficient as function of the molarity rather than the

molality. Doing so, a linear trend is obtained, as shown in Figure 4. 3.

Figure 4. 3 Molarity (mol/Lsolvent) (red line) and molality (mol/kgsolution) (blue line) as function of the logarithm of the activity coefficient of a salt. Data extracted from Ref246 and rescaled vs molarity.

Log(�̅�𝐿𝑖+𝐹−) can thus be expressed according to Equation 4. 8.

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 131

Equation 4. 8:

Log(�̅�𝐿𝑖+𝐹−) = 3.15 ∙ [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒𝑚𝑜𝑙

𝐿⁄ − 0.29

with 𝑅2 = 0.998

Besides, applying 10log function to Equation 4. 7, one can obtain Equation 4. 9.

Equation 4. 9:

log([𝐹−]𝑙𝑖𝑚) = log(𝐾𝑠) − 2 log(�̅�𝐿𝑖+𝐹−) − log ([𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒)

In Equation 4. 9, considering Ks as a constant and replacing log(�̅�𝐿𝑖+𝐹−) by its

expression described in Equation 4. 8, one can find Equation 4. 10.

Equation 4. 10:

log([𝐹−]𝑙𝑖𝑚) = C − A ∙ [𝐿𝑖+ ]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒−B ∙ log ( [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒)

where [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 is the initial concentration of Li+ in the electrolyte, A, B and C

constant values, C being proportional to Ks.

Thus, comparison between experimental data and fitted ones can be done as shown

in Figure 4. 4a and described in Equation 4. 11. Details of fitted parameters are given in

Chapter Materials and Methods.

Figure 4. 4 Logarithm of the limit of solubility of fluoride as function of initial concentration [Li+]electrolyte (mol/L) in the electrolyte. (a) Comparison between experimental data (black cross) and combination fit (orange). (b) Comparison between three fits: linear fitting (blue), logarithmic fitting (green), combination (orange).

132 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Equation 4. 11:

𝐿𝑜𝑔([𝐹−]𝑙𝑖𝑚) = −0.206 ∙ [𝐿𝑖+ ]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 −0.827 log([𝐿𝑖+ ]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒) −2.428

First, fitted values and experimental data are in good agreement, with interception

of their standard deviation range. Besides, Ks that is related to the intercept of the

equation can be determined to be: 𝐾𝑠𝑓𝑖𝑡 = 10−2.428 = 3.7 ∙ 10−3 . This value is

consistent with Ks calculated by approximating the activity coefficients (𝑎𝐿𝑖+ and 𝑎𝐹−)

by [Li+]lim and [F-]lim and taking into account LiF solubility in pure water (0.93 g/mol i.e.

3.7 ∙ 10-2 mol /L) i.e. 𝐾𝑠𝑑𝑖𝑙𝑢𝑡𝑒 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 = [𝐹−]𝑙𝑖𝑚2 ≈ 1.4 ∙ 10−3.

Nonetheless, although the theoretical relationships can be simplified as a

combination of [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 and log [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒, pure linear (see Equation 4. 12)

or pure logarithmic fits (see Equation 4. 13) as shown in Figure 4. 4b also give good

results with the interception of their standard deviation range and good assessment of

Ks value. Therefore, we must acknowledge that determining an accurate equation for

the expression of the solubility limit as a function of the electrolyte molarity must be

trickier than these simple calculations.

Equation 4. 12:

𝐿𝑜𝑔([𝐹−]𝑙𝑖𝑚) = −0.366 ∙ [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 −2.274

𝐾𝑠𝑙𝑖𝑛𝑒𝑎𝑟 𝑓𝑖𝑡 = 10−2.274 = 5.3 ∙ 10−3

Equation 4. 13:

𝐿𝑜𝑔([𝐹−]𝑙𝑖𝑚) = −1.854 ∙ log ([𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒) −2.6426

𝐾𝑠𝑙𝑜𝑔𝑎𝑟𝑖𝑡ℎ𝑚𝑖𝑐 𝑓𝑖𝑡 = 10−2.6426 = 2.3 ∙ 10−3

Using Li/LiF-coating to mimic inorganic-based SEI. Exposure to atmosphere environment, aqueous superconcentrated electrolyte and comparison with the behavior observed in organic electrolyte3

Having established that the LiF solubility limit is drastically decreased in WiSE, we

focused our attention on the ability for LiF to protect a negative electrode against

3 E-SEM images were performed by Marie-Francine Lagadec from Chaire chimie du solide et de l’énergie, Collège de France, Paris. Gustavo M. Hobold and Betar Gallant from the Department of mechanical engineering, MIT, Cambridge, carried out Li/LiF SEM images and GC-TCD/FID experiments.

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 133

reacting with water. To do so, a LiF artificial coating was prepared to mimic the native

LiF-rich SEI formed on the surface of anode materials in WiSE. Briefly, during the Li/LiF

sample preparation, the metallic Li is rolled on a current collector and then exposed to

NF3 gas at 175 °C, just below the melting temperature of Li, under mild conditions

(further details are provided in Chapter Materials and Methods). As shown in Figure 4.

5a by cross section view, the LiF layer thickness is estimated to be ≈ 30 nm. Such a LiF

thickness may be considered as representative of a native SEI. Indeed, as observed by

SEM and TEM experiments111,122,247,248, the thickness of a native LiF-based SEI is ≈ 10 nm.

Moreover, as previously demonstrated249 and observed in Figure 4. 5b and c,

homogenous and conformal deposits are realized. Thus a complete LiF coating is formed

and protects the metallic Li when further exposed to moisture or electrolytes.

Figure 4. 5 Scanning electron microscopy (SEM) images (a) Cross section SEM image of a Li/LiF pristine sample. Top-view SEM images of a Li/LiF pristine sample, (b) high magnification, (c) low magnification.

The thickness of the LiF coating layer as observed by SEM was compared with the

amount of [F-] measured after full dissolution of the LiF layer. Doing so, agreement is

met with a thickness of 30 nm, with a deviation of less than 1 %, as described in Table

4.2 (Calculation and protocol details are provided in Chapter Materials and Methods).

134 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Table 4. 2 Fluoride concentration measurement after complete dissolution of a Li/LiF sample. Comparison with expected concentration based on SEM observation. Assumptions are the following: Li/LiF sample with 30 nm thickness, 1.27 cm diameter.

Eventually, owing to this nanoscale thickness, the LiF layer cannot be detected by

laboratory XRD. However, previous demonstration was made by our collaborators (see

He et al.249) that thicker layers grown using a similar methodology but with longer

reaction times are polycrystalline.

Besides, Figure 4. 6 compares the solubility limit of LiF in aqueous and organic

electrolytes with the concentration of fluoride calculated for the complete dissolution

of the 30 nm conformal LiF-layer in 250 µL of electrolyte, i.e. a concentration of 2.02 ∙

10-2 g/L. Doing so, one can infer that in diluted aqueous electrolytes, the conformal LiF-

layer would be able to dissolve, unlike in superconcentrated aqueous electrolyte and

(1,3-dioxolane:dimethoxyethane) DOL:DME or EC:DMC-based electrolyte. This trend is

also preserved at 55°C since the LiF solubility increases by less than a factor of 2 (3.7 ±

0.4 ∙ 10-3 g/L in WiSE and 2.9 ∙ 10-3 g/L in WiBS at 55 °C), so the very limited dissolution

still enables practical cycling conditions. This implies that the instability of the LiF-based

SEI -either native or artificial- in WiSE should not arise from the complete dissolution of

the passivating layer.

Volume of the solution in

which sample is dissolved (mL)

[F-] expected

(mg/L)

[F-] measured

by ISE (mg/L)

Relative deviation

(%)

Temperature (°C)

pH of the solution

125 0.0803 0.0805 0.25 19 5.79

150 0.0669 0.0671 0.3 19.3 5.74

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 135

Figure 4. 6 Comparison of the LiF solubility limits measured in different electrolytes: water, 1 m LiTFSI, 20 m LiTFSI, 20 m LiTFSI : 8 m LiBETI, EC:DMC (1:1 vol%), DOL:DME (1:1 vol%). The reference lines indicate the fluoride concentration expected from the complete dissolution of a 0.9 cm diameter and 30 nm thickness layer (dark purple) or a 10 nm thickness (light purple) LiF layer dissolved in 250 µL of electrolyte.

Having established that LiF solubility drastically decreased in WiSE, environmental

scanning electron microscopy (E-SEM) was used to assess the protective power of the

LiF layer against moisture (see Figure 4. 7). In this experiment, the sample is kept at 20

°C while gradually increasing the relative humidity (RH) of the atmosphere in the

chamber from 0 to 90 % RH as illustrated in Figure 4. 7b.

As depicted in Figure 4. 7a, round-shaped particles start forming on the surface of a

metallic Li sample as the chamber’s RH exceeds 0 %, their occurrence increasing with

the increasing RH. On the contrary the LiF-conformal layer protects the metallic Li from

reacting with gaseous water below a threshold of 60 % RH. However, approaching water

condensation (RH ≈ 90 %), the LiF-conformal layer no longer plays a protective role and

similar (but larger) round-shaped particles are observed alike for bare metallic Li.

Interestingly, as observed in Figure 4. 7a and illustrated in Figure 4. 7c, reactivity can

first be observed at cracks and pits formed in the LiF layer, suggesting that

microstructural defects inherently present in the artificial SEI lead to reactions of the

136 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

underlying Li layer with moisture. However, testing Li/LiF samples resistance to moisture

is still far from the potential impact of aqueous superconcentrated electrolyte when in

contact with Li/LiF, therefore Li/LiF behavior in aqueous superconcentrated electrolyte

was then studied.

Figure 4. 7 (a) E-SEM images of bare Li and Li/LiF samples exposed to moisture. (b) Phase diagram for pure water, illustrative scheme of the E-SEM operation. (c) Scheme of an ideal and a real SEI exposed to moisture.

Our attention then turned to the chemical stability of the LiF-coated metallic Li

samples in both superconcentrated electrolytes, 20 m LiTFSI and 20 m LiTFSI : 8 m LiBETI.

For that, Li/LiF samples were exposed to various electrolytes while monitoring the gas

evolution by gas chromatography with thermal conductivity/flame ionization detectors

(GC-TCD/FID), as shown in Figure 4. 8.

First, for comparative purposes, we determine the gas evolution when a Li/LiF sample

is exposed to organic LP30 electrolyte for which the total amount of gas released (Figure

4. 8a) originates from the decomposition of carbonates, i.e. CO, CO2, C2H4 and CH4 as

shown in Figure 4. 8b. It stabilizes after 30 minutes at a very low value of approximately

0.5 ∙ 10-2 % in the headspace (the rest being argon), which is close to the detection limit

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 137

of the GC-FID instrumentation. Hence, even though minute cracks exist in the LiF-

conformal layer, the exposition of bare Li to LP30 eventually leads to the formation of

some polymeric/oligomeric structures that passivate the electrode250,251 and prevent

further gas evolution, explaining the decrease of gas detected over time (inset in Figure

4. 8a).

Figure 4. 8 (a) Gas evolution as a function of time for Li/LiF samples during exposure to 1M LiPF6 in EC:DMC (LP30). The gas evolution is the sum of all gases detected in (b). (b) Concentration in Carbon monoxide (dark brown), ethylene (light brown), acetylene (grey) and methane (flesh color) when exposing Li/LiF or bare metallic Li samples to LP30 electrolyte during 15 min. Hydrogen evolution as a function of time for Li/LiF samples upon exposure to (c) 20 m LiTFSI and (d) 20 m LiTFSI : 8 m LiBETI.

In contrast, a large amount of hydrogen (≈ 8 %) of the sampled gas is detected with a

LiF-protected metallic Li sample exposed to 20 m LiTFSI electrolyte (see Figure 4. 8c).

This amount then stabilizes at ≈ 4% before vanishing after 100 minutes when almost all

the metallic Li is consumed. Indeed, as described in Table 4. 3, a hydrogen molar ratio

of 26.35 % is detected after 120 min, which corresponds to 182 µmol of hydrogen

produced and thus ≈ 80% of total Li consumption (calculation parameters are given in

Table M.M. 5 in Chapter Materials and Methods). Similarly, when exposing the Li/LiF

sample to 20 m LiTFSI : 8 m LiBETI, hydrogen is detected with a concentration of ≈ 1 %.

138 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Unlike for the 20 m LiTFSI solution, this concentration remains constant throughout the

measurement, leading to a consumption of ≈ 22 % of metallic Li after 2 hours. Eventually,

such continuous hydrogen evolution indicates that, unlike for carbonate-based

electrolytes, no self-passivation occurs in superconcentrated aqueous electrolytes.

Table 4. 3 Consumption of Li after exposure to 20 m LiTFSI (WiSE) or 20 m LiTFSI : 8 m LiBETI for the Li/LiF samples.

Altogether these GC-TCD results indicate that aqueous electrolyte reaches the

underlying metallic Li electrode through microstructural defects, i.e. cracks or grain

boundaries, which have previously being observed for native SEI formed in

superconcentrated aqueous electrolytes by SEM and TEM experiments111,122,247,248.

Moreover, kinetics for water reduction is not drastically impacted by the salt

concentration7 and the amount of water contained in 250 µL of aqueous

superconcentrated electrolytes does not limit the reaction (0.489 mmol of metallic Li

available per Li/LiF sample compared to 3.49 mmol of water, see Chapter Material and

Methods for detailed calculations). Thus, one can hypothesize that the greater the

viscosity of the electrolyte, (ηWiBS = 203 mPa at 30 °C 58 and ηWiSE = 36.2 mPa at 25 °C111),

the slower the electrolyte penetrates through minute cracks present in the LiF layer,

thus explaining the differences of Li-water reactivity between the two

superconcentrated aqueous electrolytes.

Comparison of LiF behavior with Al2O3-coated Li sample4

One legitimate question arising from this study regards the quality of the LiF layer.

We thus compared the protective power of our LiF layer with that of a conformal Al2O3

layer prepared by atomic layer deposition (ALD), previously proposed to allow for

4 Li/Al2O3 coatings were provided by Christophe Lethien and Maxime Hallot from the University of Lille. GC-TCD experiments were done by Gustavo M. Hobold and Betar Gallant from the Department of mechanical engineering, MIT, Cambridge.

Li/LiF pristine exposed to

Molar ratio of H2 (after 120

min) xHydrogen(%)

Amount of H2 released (after

120 min) nHygrogen (µmol)

Ratio of Li consumed

(after 120 min) (%)

20 m LiTFSI 26.35 182 74

20 m LiTFSI : 8 m LiBETI 7.63 53 22

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 139

passivation of anode materials in WiSE as Al2O3 is not soluble in water205,207. Using a

similar approach combining E-SEM (see Figure 4. 9) and GC-TCD (see Figure 4. 10),

evolution of hydrogen upon consumption of the underlying Li electrode is once again

observed as summarized in Table 4. 4.

Table 4. 4 Consumption of Li after exposure of Al2O3-coated metallic Li exposed to 20 m LiTFSI or 20m LiTFSI : 8 m LiBETI.

Molar ratio of H2 (after 120min)

xHydrogen(%)

Amount of H2 released (after

120min) nHygrogen (µmol)

Ratio of Li consumed (after

120min) (%)

Al2O3-coated metallic Li exposed to 20m

LiTFSI 1.15 7.933 18

Al2O3-coated metallic Li exposed to 20m LiTFSI : 8m LiBETI

0.416 2.867 6.4

Environmental-SEM observations were first performed on 2 nm and 10 nm thick Al2O3

layers prepared by ALD on the surface of metallic Li, bearing in mind that the native-SEI

growing in WiSE is itself in the 10 nm thickness range. Figure 4. 9 shows the images

obtained with increasing RH. As seen before exposure to moisture, Al2O3 layers prepared

by ALD forms a granular and textured coating on the surface of Li, alike the

microstructure previously reported elsewhere for a similar coating207. Nevertheless,

when gradually increasing the water partial pressure from 0 to 90 % RH, obvious

degradations are observed starting at 30 % RH for the 2 nm thick coating (see Figure 4.

9, top row). Bearing in mind that Al2O3 does not dissolve in pure water, this change in

microstructure from granular to a cauliflower-like may arise from two effects. First,

Al2O3 can gradually transform to Al(OH)3, this phase transformation inducing a change

in molar volume (from 403 mol/cm3 for Al2O3 to 188.8 mol/cm3 for Al(OH)3, i.e. a

contraction of 7 % in volume). Second, the granular morphology of the Al2O3 coating

itself can induce reactivity of the underlying Li electrode associated with the formation

of LiOH and/or LiOH.H2O upon reaction with water. In other words, the granular

morphology of the 2 nm Al2O3 coating prepared by ALD may not prevent water from

accessing Li. Therefore, a thicker coating of 10 nm was then prepared, but similar

degradation was observed, with a cauliflower-like structure being formed upon

increased relative humidity (see Figure 4. 9, bottom row).

140 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Figure 4. 9 E-SEM images for 2 nm (top) and 10 nm (bottom) Al2O3 coated metallic Li taken successively at 0, 30, 60 and 90 % RH with zoom-ins.

To further understand the reactivity of Al2O3-coated metallic Li in contact with

superconcentrated aqueous electrolytes, GC-TCD measurements were performed alike

the ones carried out for the Li/LiF-protected samples. Upon exposure to 20 m LiTFSI, a

constant evolution of hydrogen is observed during two hours with a concentration of ≈

0.15 % of hydrogen in the headspace (see Figure 4. 10). While this concentration is much

smaller than the one previously observed with the Li/LiF sample (see Figure 4. 8c), once

normalized by the amount of Li -taking into account the surface and the thickness of the

metallic Li electrode- one can estimate that 18 % of the metallic Li was consumed by

the reaction. Thus, as observed by environmental-SEM, the conformal coating of Al2O3

does not prevent metallic Li from reacting with WiSE aqueous electrolyte but slows

down the reaction. Finally, alike for Li/LiF samples, upon exposure to 20 m LiTFSI : 8 m

LiBETI electrolyte, the concentration of hydrogen measured by GC-TCD is two to three

times lower than that measured in 20 m LiTFSI electrolyte. Nevertheless, the

concentration of gas keeps increasing upon measurement, indicating a continuous

degradation of the coating and a greater reactivity of the underlying Li electrode.

Eventually, 6.4 % of metallic Li is estimated to be consumed after two hours, as

summarized in Table 4. 4.

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 141

Figure 4. 10 Gas evolution as a function of time for 10 nm Al2O3-coated metallic Li samples after exposure to 20m LiTFSI (red) and 20m LiTFSI : 8m LiBETI (pink) aqueous superconcentrated electrolytes.

To summarize, acknowledging that the high-quality LiF layer prepared in this work,

as well as Al2O3 prepared by ALD deposition method, will always exhibit some degree of

structural defects (cracks, microporosity or else) alike a native SEI which consists of a

mosaic of LiF grains111,240, our study highlights that even if the use of superconcentrated

aqueous electrolytes prevents the dissolution of the inorganic SEI compounds, these

electrolytes are deprived of self-passivating ability through the formation of an organic-

inorganic SEI outer layer required to stabilize anode materials.

142 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Filling the structural defects by presoaking in organic electrolyte: assessment of the importance of an organic-inorganic based SEI5

To overcome the absence of self-passivation, Li/LiF samples were soaked in three

organic-based solutions: pure fluoro-ethylene carbonate (FEC) solvent and two

electrolytes, namely 7 M LiFSI in FEC and 2 M LiFSI : 1 M LiTFSI in DOL:DME (1:1 %vol)

with 3 % LiNO3 additive, both known for forming good elastomeric passivating SEI with

a LiF-rich inner layer and enabling very high Coulombic efficiencies for Li

plating/striping252–254. After soaking, during which any possible defects within the LiF

layer may be further passivated by the formation of an additional organic-inorganic layer

derived from a non-aqueous electrolyte, the samples are exposed to superconcentrated

aqueous electrolyte and hydrogen evolution in measured by GC-TCD (Figure 4. 11).

Figure 4. 11 Illustration of the presoaking step in organic solvent/electrolyte, further exposed to aqueous superconcentrated electrolyte.

Hydrogen concentration is measured over time for pristine Li/LiF sample and samples

presoaked in 50 µL or 2 mL of FEC further exposed to WiBS (see Figure 4. 12a). First, the

5 Gustavo M. Hobold did GC-TCD experiments from the Department of mechanical engineering, MIT, Cambridge. Rui Guo, Gustavo M. Hobold and Betar Gallant from the Department of mechanical engineering, MIT, Cambridge, carried out XPS analysis.

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 143

hydrogen concentration measured after presoaking the sample in 50 µL of pure FEC is

similar than for the pristine LiF-protected Li sample (22 % of metallic Li consumption for

pristine sample vs 19 % for the sample presoaked in 50 µL FEC, see Table 4. 5).

Interestingly, when increasing the volume of FEC during the soaking step (from 50 µL to

2 mL), we observe an increase of hydrogen evolution (≈ 3 % after 15 min compared to ≈

1 % for the pristine Li/LiF as shown in Figure 4. 12a) which is translated into an increased

metallic Li consumption from 22 % to 55 % (see Table 4. 5). Combining this observation

with post soaking XPS analysis (see Figure 4. 12b and c), one can attribute it to the partial

dissolution of the LiF layer in FEC as the intensity of both the F 1s and the Li 1s signals

decreases for the sample presoaked in 2 mL FEC (see signal [1] on Figure 4. 12b and c).

Figure 4. 12 (a) Hydrogen evolution as function of time for pristine Li/LiF sample (pink), presoaked in 2 mL FEC (light blue, empty square), presoaked in 50 µL FEC (light blue, full square) prior to exposure to 20 m LiTFSI : 8 m LiBETI. X-ray photoelectron spectroscopy (XPS) spectrum of (b) F 1s and (c) Li 1s of pristine Li/LiF [0], presoaked Li/LiF sample in 2 mL FEC [1], presoaked Li/LiF sample in 50 µL FEC [2].

144 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Table 4. 5 Consumption of metallic Li after exposure to 20 m LiTFSI : 8 m LiBETI for the Li/LiF sample: pristine, presoaked in 2 mL FEC, presoaked in 50 µL FEC assuming hydrogen evolution only caused by HER in Li consumption ratio calculation.

Unfortunately, the solubility of the LiF layer in FEC as observed during these

experiments could not be correlated with a greater solubility limit for LiF in FEC, as

determined in Section 0 by ISE methods for other solvents. Indeed, the main advantage

of the fluoride ion selective electrode technique lies in its ability to measure fluoride

concentration in organic solvent/electrolyte diluted in an aqueous ionic strength buffer.

Hence, Strmcnik et al.244 showed that by diluting LP57 (EC/EMC/LiPF6) in ionic aqueous

strength buffer by a factor 100, the hydrolysis of PF6- anions was quenched.

Nonetheless, despite this approach shows good results in LP57, continuous hydrolysis

of FEC was observed in the ionic aqueous strength buffer with a continuous increase of

the probed [F-] concentration for over a week (from ≈ 3 g / L at 23 °C to ≈ 180 g /L at 23

°C for both LiF-saturated or non-LiF-saturated FEC solution after one week). This effect

prevented us to determine the solubility limit of fluoride in pure FEC.

To overcome the dissolution of the LiF coating, samples were then soaked in an ether-

based electrolyte (1 M LiTFSI : 2 M LiFSI in DOL:DME + 3 % LiNO3). As mentioned in

section 4.1 and shown in Figure 4. 2, DOL:DME-based electrolyte should not dissolve LiF

layer. When exposing this sample to WiBS (see Figure 4. 13a, purple dots), the hydrogen

evolution is greater after 15 min (≈ 1.6 %) than for the pristine Li/LiF sample.

Nevertheless, the signal rapidly decreases and stabilizes below 1 % of hydrogen detected

in the headspace. Overall, after this equilibration period, soaking the Li/LiF anode in 1 M

LiTFSI : 2 M LiFSI in DOL:DME + 3% LiNO3 has a slight positive impact on the gassing,

decreasing the metallic Li consumption through hydrogen evolution by 4 % when

compared with the pristine Li/LiF sample (see Table 4. 6).

Exposure to 20 m LiTFSI : 8 m LiBETI

Molar ratio of H2 (after 120min)

xHydrogen(%)

Amount of H2 released (after

120min) nHydrogen (µmol)

Ratio of Li consumed (after

120min) (%)

Li/LiF pristine 7.63 53 22

Li/LiF presoaked in 2 mL pure FEC

19.6 135 55

Li/LiF presoaked in 50 µL pure FEC

6.85 47 19

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 145

To further study the impact of presoaking in organic electrolyte, we then soaked a

Li/LiF sample in an organic superconcentrated electrolyte, namely 7 M LiFSI in FEC,

before exposing it to WiBS (see Figure 4. 13a, yellow dots). This soaking step is found to

have a beneficial impact and diminishes the hydrogen evolution upon exposure to WiBS,

which approaches the detection limit even at the earlier stage of exposure (0.035 % of

the total headspace after 15 min). Furthermore, even if the amount of hydrogen

detected slightly increases with time, it corresponds to a consumption of metallic Li of

only 4 % after two hours considering HER as the sole source for H2 evolution, compared

to 22 % for the pristine Li/LiF sample as described in Table 4. 6.

Figure 4. 13 Hydrogen evolution as function of time for Li/LiF samples. Li/LiF presoaked in pure FEC (light blue), Li/LiF presoaked in 7 M LiFSI in FEC (yellow) or presoaked in 2 M LiFSI : 1 M LiTFSI in DOL:DME (purple) prior to exposure to (a) 20 m LiTFSI : 8 m LiBETI with pristine Li/LiF in pink, (b) 20 m LiTFSI, with pristine Li/LiF in red.

Table 4. 6 Consumption of Li after exposure to 20 m LiTFSI : 8 m LiBETI for the Li/LiF pristine, presoaked in 2 mL FEC, presoaked in 50 µL FEC, presoaked in 7 M LiFSI in FEC or 2 M LiFSI : 1 M LiTFSI in DOL:DME, 3 % LiNO3 assuming hydrogen evolution only caused by HER in metallic Li consumption ratio calculation.

Besides, similar experiments have been performed in WiSE (see Figure 4. 13b). As in

WiBS, the presoaking in 1 M LiTFSI : 2 M LiFSI in DOL:DME + 3 % LiNO3 and in 7 M LiFSI

Exposure to 20 m LiTFSI : 8 m LiBETI

Molar ratio of H2 (after 120min)

xHydrogen(%)

Amount of H2 released (after

120min) nHydrogen (µmol)

Ratio of Li consumed

(after 120min) (%)

Li/LiF pristine 7.63 53 22

Li/LiF presoaked in 50 µL of 2 M LiFSI + 1 M LiTFSI in DOL:DME, 3 % LiNO3

6.44 44 18

Li/LiF presoaked in 50 µL of 7 M LiFSI in FEC

1.41 10 4

146 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

FEC has a beneficial, but reduced, impact on the hydrogen gassing. The metallic Li

consumption of LiF-coated Li samples exposed to WiSE fades from 74 % for the pristine

sample to 27 % in 1 M LiTFSI : 2 M LiFSI in DOL:DME and 33 % in 7 M LiFSI in FEC,

respectively, as described in Table 4. 7. Furthermore, trends observed for both organic

electrolytes are similar than in WiBS. In details, after 15 min exposure to WiSE, the

hydrogen concentration measured for the sample soaked in 1 M LiTFSI : 2 M LiFSI in

DOL:DME increases to ≈ 3 % then stabilizes at ≈ 1 % while it is mostly constant during

the two hours experiment after soaking in 7 M LiFSI in FEC. However, one can notice

that although the decrease of metallic Li consumption is significant, the differences

recorded between these two soaking steps is less remarkable than in WiBS and the

metallic Li consumption is still greater than the one of the sample exposed to WiBS. One

possible explanation would be the difference of kinetics for such electrolyte penetration

being dependent on its viscosity and the mass transport across the microporosity and/or

cracks leading to a faster access of water in the case of WiSE and preventing from greater

performances.

Table 4. 7 Consumption of metallic Li after exposure to 20 m LiTFSI for the Li/LiF samples: pristine, presoaked in 7 M LiFSI in FEC or 2 M LiFSI : 1 M LiTFSI in DOL:DME, 3 % LiNO3 assuming HER as sole source for hydrogen evolution in metallic Li consumption ratio calculation.

Exposure to 20 m LiTFSI

Molar ratio of H2 (after 120 min)

xHydrogen(%)

Amount of H2 released (after 120

min) nHygrogen (µmol)

Ratio of Li consumed (after

120 min) (%)

Li/LiF pristine 26.35 182 74

Li/LiF presoaked in 7 M LiFSI in FEC

11.53 80 33

Li/LiF presoaked in 2 M LiFSI : 1 M LiTFSI in DOL:DME, 3 % LiNO3

7.94 67 27

Based on the decrease of hydrogen gassing observed during GC-TCD measurements,

XPS analysis were performed to analyze the surface of the Li/LiF samples after the

presoaking step. XPS spectrum of C 1s, O 1s, F 1s, Li 1s, N 1s and S 2p are shown in Figure

4. 14. First of all, presoaking step in DOL:DME-based electrolyte or FEC-based electrolyte

almost fully covered the Li/LiF layer with another layer. Indeed, the signal intensity of

LiF is barely visible in F 1s and Li 1S spectrum (Figure 4. 14c and d). Moreover, the nature

of the additional covering layer formed during presoaking step may be attributed to the

formation of a salt-derived layer as shown by LiTFSI or LiFSI signals in Figure 4. 14b, c, e

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 147

and f. Regarding presoaking step in DOL:DME-based electrolyte, C 1s spectra gives a

greater amount of carbonaceous compounds, i.e. more organic-based species deposited

on the Li/LiF sample that can be due to the presence of C-F bound in TFSI salt. Therefore,

one can suspect that the beneficial impact of 7 M LiFSI in FEC and 1 M LiTFSI : 2 M LiFSI

in DOL:DME + 3 % LiNO3 arises from their ability to form an inorganic-organic SEI upon

decomposition on the surface of Li, 71,249 both being able to partially compensate the

microstructural defects of the ex-situ LiF interface. Hence, post-soaking XPS analysis

reveal the formation of a salt-derived inorganic SEI for both electrolytes with an organic

contribution for the DOL:DME-based electrolyte.

Consequently, the beneficial impact on hydrogen gassing when a Li/LiF sample is

exposed to WiBS ranks as follows: pristine Li/LiF < pure FEC < 1 M LiTFSI : 2 M LiFSI in

DOL:DME + 3% LiNO3 < 7 M LiFSI in FEC. From this trend, one can conclude that the

ability of the electrolyte to form and maintain stable a LiF-rich SEI inner layer is of prime

importance to positively stabilize the interface. Additionally, having an organic

contribution to the SEI, as we observe for DOL:DME tends to be beneficial. All of this

without promoting the partial dissolution of the SEI.

148 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Figure 4. 14 X-ray photoelectron spectroscopy (XPS) spectrum of (a) C 1s, (b) O 1s, (c) F 1s, (d) Li 1s, (e) N 1s, (f) S 2p of pristine Li/LiF [0], presoaked in 50 µL FEC [1], presoaked in 7 M LiFSI [2] and presoaked in 2 M LiFSI : 1 M LiTFSI in DOL:DME, 3% LiNO3 [3].

Conclusion of the chapter

In this Chapter, we first determined the fluoride solubility limit of LiF salt as function

of LiTFSI salt concentration from diluted aqueous media to aqueous superconcentrated

electrolyte. Based on these measurements, one can conclude that LiF-based SEIs is non-

soluble in both 20 m LiTFSI and 20 m LiTFSI : 8 m LiBETI. Therefore, LiF coatings onto

metallic Li were used to mimic LiF-based SEI in aqueous superconcentrated electrolyte.

Combining E-SEM with GC-TCD measurements during Li/LiF samples exposition to

moisture or WiSE/WiBS, respectively, enable to find that unlike in classical organic

electrolyte, aqueous superconcentrated electrolyte cannot self-passivate the interface.

These results are further confirmed by testing insoluble Al2O3 coating resistance to

WiSE/WiBS exposure. Indeed, hydrogen production is also found even though less

intense. Therefore, a presoaking step in organic electrolyte known to form good

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 149

passivating SEI with LiF-rich inner layer and organic outer layer was employed to analyze

the importance of organic-inorganic species in the SEI to maintain it stable. To do so, 1

M LiTFSI : 2 M LiFSI in DOL:DME + 3 % LiNO3 or 7 M LiFSI in FEC were tested. GC-TCD

results show that hydrogen evolution is diminished. This may be explained by the

formation of a salt-derived SEI, as shown by XPS post-soaking analysis, as well as the

presence of organic compounds in the case of the DOL:DME-based electrolyte.

Moreover, combining these analysis and comparing them with those obtained in organic

electrolytes, we conclude on the partially-protective role of both native and artificial LiF-

rich SEI for superconcentrated aqueous electrolytes. More importantly, we demonstrate

that a salt-derived inorganic and solvent-derived organic layers play a crucial role in the

SEI’s ability to self-repair and allow for cycling anode material outside of the

thermodynamic stability window of the electrolyte.

Indeed, as recently demonstrated with the use of organic co-solvents such as

acetonitrile 192, DMC191,255 or ether-based one (TEGDME)193, organic SEI outer layer

seems to be beneficial. Similarly, additives such as urea190,256 were also shown to have

positive effects on the cycling behavior of Li4Ti5O12 anode (1.7 V vs Li+/Li) in WiSE-based

batteries owing to its decomposition in both inorganic (Li2CO3) and amorphous organic

(polyuria) layers. Nevertheless, bearing in mind that in these electrolytes, the first

solvation sheath of Li+ contains both water and organic species, it is tempting to

conclude that both should be simultaneously present at the interface, leading to the

detrimental reduction of water simultaneously to the SEI formation (knowing that

solvating water molecules are more reactive than non-solvating ones144). Moreover,

despite the diminished hydrogen evolution with an artificial inorganic LiF layer

presoaked in 7 M LiFSI in FEC, these ex-situ protections are not sufficient to envision

practical applications. Indeed, considering the electrolyte’s water content as the limiting

factor, in the best-case scenario, fewer than 50 cycles in WiSE and 250 cycles in WiBS

would be achieved before the drying out of a cell cycled at 1C rate (see Table 4. 8).

Furthermore, taking into account the quantity of metallic Li used in this study (489

µmol), only six hours would be needed to fully consume the metallic Li in WiSE, and less

than fifty hours in WiBS. These simple calculations do not take into account electrolyte

crystallization but merely the full consumption of water, thus providing grossly

underestimated numbers that would be even worse in practice. Evidently, using metallic

Li as an anode leads to harsh degradation which could be partially alleviated, or at least

slowed down, by the use of an anode material with a greater redox potential (LTO, TiO2,

Mo6S8, LTP or else). Indeed, around 8750 cycles could be performed using unprotected

150 Chapter 4 – Understanding of inorganic SEI limitations in WiSE

Mo6S8 as negative electrode before the complete drying out of the cell as shown in Table

4. 8, this gain in cycling life coming at the expense of energy density.

Table 4. 8 Estimation of water and metallic Li consumption according to the gassing measured by GC-TCD.

*based on Chapter 2 and 3, rate of water consumption is calculated to be ≈ 0.0005

%initial water amount/hself-discharge for a stable SEI formed and ≈ 0.004 %initial water amount/hcycling at

1C during cycling. Therefore, in this calculation, we only consider cycling data.

As a conclusion, altogether, this work evidences that 1) the use of superconcentrated

aqueous electrolytes does not allow for the self-passivation of the SEI at the anode; 2)

the solubility is not the predominant factor governing the poor stability of such inorganic

coatings. These results are in good agreement with the performances obtained with self-

discharge protocol (drop of discharge capacity and Coulombic efficiency after resting

period) in Chapter 2 and 3. Eventually, while inorganic-rich interphases can physically

impede the reaction and allow for a certain stability over a short period, the aqueous

electrolyte will inherently reaches the underlying anode material owing to extrinsic

defects, which will be present in any practical the SEI. Furthermore, bearing in mind that

cracks and/or microporosity may continuously form upon cycling, triggering their self-

passivation is necessary to protect the underlying electrode. Towards that goal, two

strategies can be envisioned, either with additives or with organic coatings. Regarding

Case Li/LiF, cracks filled with 7M LiFSI in FEC,

exposure to WiSE

Li/LiF, cracks filled with 7M LiFSI in FEC, exposure to

WiBS

Mo6S8 electrode used in WiSE

(based on Chapter 2 and 3)

Water consumption µmo𝑙𝐻20/ℎ

80 10 *0.14

Water amount in 250 µL of electrolyte

(mmol) 3.49 2.48 3.49

Time (h) needed to consume water

44 248 25000

Number of cycle considering 1C C-rate

(1 Li+ inserted / h) <50 <250 <6300

Li amount used in the study (µmol)

489 489

Time (h) needed to consume Li

6.1 48.9

Chapter 4 – Understanding of inorganic SEI limitations in WiSE 151

the use of additives, they must be targeted following their ability to preferentially

solvate Li+ and thus displace water from its solvation sheath. Organic coatings can also

be developed by a grafting strategy or by laminating polymer protective layers, as

previously reported120,204. Towards these goals, the perspectives of this thesis will be

dedicated to polymer layer as artificial SEI for aqueous superconcentrated electrolyte

LIB application.

GENERAL CONCLUSION AND PERSPECTIVES

154 General conclusion

Through this phd thesis, the use of aqueous superconcentrated electrolyte, namely

Water-in-salt, was investigated in LIB. The attraction towards WiSE is explained by

structural and physical modifications arising from the formation of two interpenetrated

networks in these electrolytes, a water-rich network and an anion-rich one. This peculiar

solvation structure was shown to enable fast transport with high Li+ cation transference

number while promoting the formation of a salt-derived inorganic-rich SEI at the

negative interface and preventing water oxidation on the positive side thanks to the

formation of an hydrophobic double-layer. Therefore, the electrochemical stability

window of these electrolytes was reported to increase above 2 V. These changes in

physico-chemical properties and in interfacial reactivity, and the influence of

superconcentrated aqueous electrolytes on LIB performances are described in the first

chapter.

Besides, the set-up of a systematic study on both the anodic and cathodic stability of

20 m LiTFSI was performed as function of the operating conditions to assess the viability

of WiSE electrolyte in LIB. Although the impact of the increase in concentration was

shown to positively impact cell performances at room temperature, by changing the C-

rate, the presence of parasitic reaction is brought to light. These parasitic reactions were

found to be even further damaging at higher temperature. Therefore, operando gas

monitoring experiments were carried out to determine the nature of the parasitic

reactions causing performances decay. Combining pressure cell and online

electrochemical mass spectrometry, hydrogen evolution (HER) was found to occur both

during cycling and resting period, triggering a so-called self-discharge. However, one

must note that self-discharge appears partially reversible in terms of lithium balance,

the major drawback being the irreversible water consumption. Eventually, this study

was completed with the use of 20 m LiTFSI : 8 m LiBETI to illustrate that the

aforementioned limitations are general to aqueous superconcentrated electrolytes used

in LIB. Indeed, alike what was observed in 20 m LiTFSI, this study show that the HER

takes place during cycling and resting period, suggesting a weak passivation of the SEI

formed in these WiSEs, though HER intensity was found much decreased in 20 m LiTFSI

: 8 m LiBETI when compared to 20 m LiTFSI. Besides, on the anodic side, the oxygen

evolution impact was shown to be non-existent at room temperature, and to have

limited impact at higher temperature. Altogether, the results found in Chapter 2 show

that the main issue regarding the use of WiSE regards the negative electrode and the

SEI stability.

General conclusion 155

To understand these limitations, the native SEI stability was investigated by

electrochemical characterizations (cyclic voltammetry and electrochemical impedance

spectroscopy), showing the passivating power of the SEI but its poor efficiency towards

preventing water consumption. The later was further highlighted by differential

scanning calorimetry showing a shift in crystallization peak attesting of an increase in

salt concentration after cycling and resting period. Besides, rate of water consumption

during resting was found in the same order of magnitude than the rate of water

consumption albeit smaller. Moreover, the de-intercalation rates of Li+ during self-

discharge and cycling were found to largely differ. To comprehend this observation, the

activation energy of self-discharge phenomenon occurring during resting was

determined and compared to the activation energy of the HER taking place during

cycling. Experimental results confirmed a similar activation energy (≈ 20 kJ/mol),

suggesting that self-discharge is governed by HER rather than by delithiation of the

negative electrode. Altogether, in Chapter 3 the dramatic impact of water reduction

during self-discharge and cycling is highlighted when relying on the formation of a native

SEI.

To understand the relative instability of the native SEI, the solubility of LiF (the major

contributor to the SEI) was studied in several media, from pure water, 1 m LiTFSI to 20

m LiTFSI and 20 m LiTFSI : 8 m LiBETI WiSEs. Experimental results demonstrate that a

≈ 10 nm LiF thickness (comparable to a native SEI one) should not completely dissolve

in aqueous superconcentrated electrolyte, showing that solubility is not the issue. To

probe if the microstructure of the native SEI is limiting its stability, an artificial LiF SEI

was prepared onto metallic Li (Li/LiF) to assess the efficiency of an inorganic coating to

protect the negative electrode from water reduction. Combining E-SEM and GC-TCD

measurements, the exposure to moisture or aqueous superconcentrated electrolyte

was found to enable water to react with Li, despite the presence of a conformal LiF

coating. More impactful, WiSE or WiBS do not allow for self-passivation unlike what is

observed in organic electrolyte (here, in LP30). The absence of self-passivation was

confirmed using Al2O3 coating, known to be insoluble in water, demonstrating that in

the absence of self-passivation water reacts with the underlying electrode through

surface defects such as micro-porosity or cracks in the SEI. In order to overcome this

limitation, a presoaking step was performed in organic electrolyte known to form good

passivating SEI with LiF-rich inner layer and organic outer layer: 1 M LiTFSI : 2 M LiFSI in

DOL:DME + 3 % LiNO3 or 7 M LiFSI in FEC. GC-TCD measurements show that hydrogen

evolution is diminished when further exposing the electrode to WiSE or WiBS, owing to

the deposition of salt-derived compounds as shown by XPS. As a consequence, the use

156 General conclusion

of artificial inorganic SEI coupled with presoaking in organic electrolyte demonstrates

the need for the SEI to be able to self-repair. It further suggests that the presence of

organic component further promote SEI stability, and that negative electrode may

operate at low potential far below the ESW limit if coated with such organic layer.

The figure-of-merits, as detailed in Chapter 3, compare the performances of WiSE-

based LIB with that of other rechargeable battery technologies such as Lead-acid, Ni-Cd,

Ni-MH and commercial LIB (using diluted organic electrolyte). Evidently, WiSE-based

batteries can only operate within a ≈ 2 V operating window to avoid parasitic reactions,

thus drastically limiting the energy density and specific energy compared to Ni-MH

batteries. Moreover, while the use of chemical-engineering means to handle gas

generated during cycling could be envisioned, alike what is done for other aqueous

secondary battery (Lead-acid, etc.), the superconcentration may cause rapid

crystallization of the electrolyte for vented or flooded-type cells that evacuate gases

produced without compensating water consumption. However, paradoxically, too few

oxygen is produced at the positive while hydrogen production happens concomitantly

with Li+ insertion at the negative, thus preventing the implementation of the “oxygen

cycle” used in sealed-type batteries. Apart from energy consideration, a major drawback

of most of the aqueous superconcentrated electrolytes is their narrow thermal stability

range. Indeed, they often crystallize near room temperature. Several strategies were

previously shown to overcome the operating temperature range limitation by the use

of asymmetric anion and/or the introduction of organic co-solvent that downshift the

melting point, as detailed in the first chapter. Finally, and more critically, while self-

discharge is found reversible in terms of Li+ balance, it is responsible for irreversible

water consumption. Therefore, means need to be found to reduce capacity fading and

prevent electrolyte crystallization for large-scale commercialization.

As a summary, WiSE-based batteries still face many drawbacks that must be

overcome before to attain satisfactory performance when compared with commercial

LIB batteries. Many strategies are envisioned to improve these systems. As detailed in

the first chapter, the introduction of organic-solvent was brought to tune the first

solvation sheath of Li+ promoting the organic co-solvent reduction and thus its

participation to SEI formation while (i) reducing the amount of water in the first

solvation sheath and (ii) disrupting the water H-bonds network. Nonetheless, bearing in

mind that water is still partially involved in the solvation sheath, it is tempting to

conclude that both the organic co-solvent and water will be reduce at low potential,

thus not fully addressing the cathodic challenge. Based on the performances first

General conclusion 157

reported in 2017 by Yang et al.204, the use of organic polymer coating could be seen as

a promising path to develop. Therefore, we tried to reproduce their results and to

assemble a battery, as described in Figure Conclusion. 1. The polymer coating is based

on 0.5 M LiTFSI dissolved in a highly fluorinated ether solvent (HFE): DMC mixture (95:5

vol%) with 10 wt% PEO. Here, DMC is used as a co-solvent to promote LiTFSI and PEO

dissolution in HFE. It is then removed during the fabrication process and does not

participate to the final coating layer. However, during our trials, difficulties to dissolve

LiTFSI and PEO in such low quantity of DMC were encountered, forcing us to move to

acetonitrile as co-solvent to enable the dissolution201. Once the gel is obtained, the

deposition on the negative electrode was found tricky/hard to reproduce. Indeed, the

gel is sticky and need to be heated up to flow in liquid state during coating procedure.

Therefore, no success in cycling such polymer-protected WiSE batteries was obtained

yet, but trials are being performed. As a summary, even though the use of polymer

coating was reported to enable the cycling of metallic Li or graphite, to the best of my

knowledge, its technical implementation requires improvement.

Figure Conclusion. 1 Scheme of a Li-ion battery relying on the use of a polymer coating on the negative electrode. The polymer coating is based on 0.5 M LiTFSI in a highly fluorinated ether (HFE):DMC (95:5 vol%) mixture with 10 wt% polyethylene oxide (PEO).

REFERENCES

(1) International energy agency. World Energy Outlook 2020. 2020.

(2) International energy agency. World Energy Outlook 2021. 2021, 386.

(3) Bloomberg New Energy Finance, “Electric Vehicle Outlook 2020”,

BloombergNEF, New-York. 2020.

(4) Avicenne. Avicenne Energy Report: EU Battery Demand and Supply (2019-2030)

in a Global Context.

(5) Cano, Z. P.; Banham, D.; Ye, S.; Hintennach, A.; Lu, J.; Fowler, M.; Chen, Z.

Batteries and Fuel Cells for Emerging Electric Vehicle Markets. Nat Energy 2018,

3 (4), 279–289. https://doi.org/10.1038/s41560-018-0108-1.

(6) May, G. J.; Davidson, A.; Monahov, B. Lead Batteries for Utility Energy Storage:

A Review. Journal of Energy Storage 2018, 15, 145–157.

https://doi.org/10.1016/j.est.2017.11.008.

(7) Reddy, T.; Linden, D. Linden’s Handbook of Batteries (4th Edition).; McGraw-

Hill Professional Publishing: New York, USA, 2010.

(8) Van den Bossche, P.; Vergels, F.; Van Mierlo, J.; Matheys, J.; Van Autenboer, W.

SUBAT: An Assessment of Sustainable Battery Technology. Journal of Power

Sources 2006, 162 (2), 913–919. https://doi.org/10.1016/j.jpowsour.2005.07.039.

(9) Beck, F.; Rüetschi, P. Rechargeable Batteries with Aqueous Electrolytes.

Electrochimica Acta 2000, 45 (15–16), 2467–2482. https://doi.org/10.1016/S0013-

4686(00)00344-3.

(10) Rouxel, J.; Danot, M.; Bichon, M. Les Composites Intercalaires NaxTiS2. Etude

Gènérale Des Phases NaxTiS2 et KxTiS2. Bulletin de la Société Chimique de France

1971, 11, 3930–3936.

(11) Whittingham, M. S. Intercalation Chemistry and Energy Storage. Journal of Solid

State Chemistry 1979, 29 (3), 303–310. https://doi.org/10.1016/0022-

4596(79)90187-7.

(12) Whittingham, M. S. Electrical Energy Storage and Intercalation Chemistry.

Science 1976, 192 (4244), 1126–1127.

https://doi.org/10.1126/science.192.4244.1126.

(13) Whittingham, M. S. CHALCOGENIDE BATTERY, US4009052A. 1977, 5.

160 References

(14) Lin, D.; Liu, Y.; Cui, Y. Reviving the Lithium Metal Anode for High-Energy

Batteries. Nature Nanotech 2017, 12 (3), 194–206.

https://doi.org/10.1038/nnano.2017.16.

(15) Blomgren, G. E. The Development and Future of Lithium Ion Batteries. J.

Electrochem. Soc. 2017, 164 (1), A5019–A5025.

https://doi.org/10.1149/2.0251701jes.

(16) Kalyani, P.; Kalaiselvi, N. Various Aspects of LiNiO 2 Chemistry: A Review.

Science and Technology of Advanced Materials 2005, 6 (6), 689–703.

https://doi.org/10.1016/j.stam.2005.06.001.

(17) Yabuuchi, N.; Ohzuku, T. Novel Lithium Insertion Material of

LiCo1/3Ni1/3Mn1/3O2 for Advanced Lithium-Ion Batteries. Journal of Power

Sources 2003, 119–121, 171–174. https://doi.org/10.1016/S0378-7753(03)00173-

3.

(18) Noh, H.-J.; Youn, S.; Yoon, C. S.; Sun, Y.-K. Comparison of the Structural and

Electrochemical Properties of Layered Li[NixCoyMnz]O2 (x = 1/3, 0.5, 0.6, 0.7, 0.8

and 0.85) Cathode Material for Lithium-Ion Batteries. Journal of Power Sources

2013, 233, 121–130. https://doi.org/10.1016/j.jpowsour.2013.01.063.

(19) Rozier, P.; Tarascon, J. M. Review—Li-Rich Layered Oxide Cathodes for Next-

Generation Li-Ion Batteries: Chances and Challenges. J. Electrochem. Soc. 2015,

162 (14), A2490–A2499. https://doi.org/10.1149/2.0111514jes.

(20) Thackeray, M. M.; David, W. I. F.; Bruce, P. G.; Goodenough, J. B. Lithium

Insertion into Manganese Spinels. Materials Research Bulletin 1983, 18 (4), 461–

472. https://doi.org/10.1016/0025-5408(83)90138-1.

(21) Lazzari, M.; Scrosati, B. A Cyclable Lithium Organic Electrolyte Cell Based on

Two Intercalation Electrodes. J. Electrochem. Soc. 1980, 127 (3), 773–774.

https://doi.org/10.1149/1.2129753.

(22) Yazami, R.; Touzain, Ph. A Reversible Graphite-Lithium Negative Electrode for

Electrochemical Generators. Journal of Power Sources 1983, 9 (3), 365–371.

https://doi.org/10.1016/0378-7753(83)87040-2.

(23) Basu, S.; Somerset, N. J. US4423125A, Bell Laboratories. 1983, 4.

(24) Fong, R.; von Sacken, U.; Dahn, J. R. Studies of Lithium Intercalation into Carbons

Using Nonaqueous Electrochemical Cells. J. Electrochem. Soc. 1990, 137 (7),

2009–2013. https://doi.org/10.1149/1.2086855.

(25) Fujimoto, M.; Yoshinaga, N.; Furukawa, N.; Nohma, T.; Takahashi, M. LITHIUM

SECONDARY BATTERY, US5686138A. 1994, 55.

(26) Yoshino, A. The Birth of the Lithium-Ion Battery. Angew. Chem. Int. Ed. 2012, 51

(24), 5798–5800. https://doi.org/10.1002/anie.201105006.

References 161

(27) Tarascon, J. M.; Armand, M. Issues and Challenges Facing Rechargeable Lithium

Batteries. Nature 2001.

(28) Xu, K. Nonaqueous Liquid Electrolytes for Lithium-Based Rechargeable Batteries.

Chem. Rev. 2004, 104 (10), 4303–4418. https://doi.org/10.1021/cr030203g.

(29) Peled, E.; Menkin, S. Review—SEI: Past, Present and Future. J. Electrochem. Soc.

2017, 164 (7), A1703–A1719. https://doi.org/10.1149/2.1441707jes.

(30) Peled, E. The Electrochemical Behavior of Alkali and Alkaline Earth Metals in

Nonaqueous Battery Systems—The Solid Electrolyte Interphase Model. J.

Electrochem. Soc. 1979, 126 (12), 2047–2051. https://doi.org/10.1149/1.2128859.

(31) Zaban, A.; Aurbach, D. Impedance Spectroscopy of Lithium and Nickel Electrodes

in Propylene Carbonate Solutions of Different Lithium Salts A Comparative Study.

Journal of Power Sources 1995, 54 (2), 289–295. https://doi.org/10.1016/0378-

7753(94)02086-I.

(32) Edström, K.; Gustafsson, T.; Thomas, J. O. The Cathode–Electrolyte Interface in

the Li-Ion Battery. Electrochimica Acta 2004, 50 (2–3), 397–403.

https://doi.org/10.1016/j.electacta.2004.03.049.

(33) Liu, Y.-M.; G. Nicolau, B.; Esbenshade, J. L.; Gewirth, A. A. Characterization of

the Cathode Electrolyte Interface in Lithium Ion Batteries by Desorption

Electrospray Ionization Mass Spectrometry. Anal. Chem. 2016, 88 (14), 7171–

7177. https://doi.org/10.1021/acs.analchem.6b01292.

(34) Bassett, K. L.; Özgür Çapraz, Ö.; Özdogru, B.; Gewirth, A. A.; Sottos, N. R.

Cathode/Electrolyte Interface-Dependent Changes in Stress and Strain in Lithium

Iron Phosphate Composite Cathodes. J. Electrochem. Soc. 2019, 166 (12), A2707–

A2714. https://doi.org/10.1149/2.1391912jes.

(35) Goodenough, J. B.; Kim, Y. Challenges for Rechargeable Li Batteries. Chem.

Mater. 2010, 22 (3), 587–603. https://doi.org/10.1021/cm901452z.

(36) Peljo, P.; Girault, H. H. Electrochemical Potential Window of Battery Electrolytes:

The HOMO–LUMO Misconception. Energy Environ. Sci. 2018, 11 (9), 2306–

2309. https://doi.org/10.1039/C8EE01286E.

(37) Chen, W.; Ambrosio, F.; Miceli, G.; Pasquarello, A. Ab Initio Electronic Structure

of Liquid Water. Phys. Rev. Lett. 2016, 117 (18), 186401.

https://doi.org/10.1103/PhysRevLett.117.186401.

(38) Borodin, O.; Self, J.; Persson, K. A.; Wang, C.; Xu, K. Uncharted Waters: Super-

Concentrated Electrolytes. Joule 2020, 4 (1), 69–100.

https://doi.org/10.1016/j.joule.2019.12.007.

162 References

(39) Yamada, Y.; Yamada, A. Review—Superconcentrated Electrolytes for Lithium

Batteries. Journal of The Electrochemical Society 2015, 162 (14), A2406–A2423.

https://doi.org/10.1149/2.0041514jes.

(40) Suo, L.; Hu, Y.-S.; Li, H.; Armand, M.; Chen, L. A New Class of Solvent-in-Salt

Electrolyte for High-Energy Rechargeable Metallic Lithium Batteries. Nat

Commun 2013, 4 (1), 1481. https://doi.org/10.1038/ncomms2513.

(41) Seo, D. M.; Boyle, P. D.; Borodin, O.; Henderson, W. A. Li+ Cation Coordination

by Acetonitrile—Insights from Crystallography. RSC Adv. 2012, 2 (21), 8014.

https://doi.org/10.1039/c2ra21290k.

(42) Kunz, W.; Barthel, J.; Klein, L.; Cartailler, T.; Turq, P.; Reindl, B. Lithium

Bromide in Acetonitrile: Thermodynamics, Theory, and Simulation. J Solution

Chem 1991, 20 (9), 875–891. https://doi.org/10.1007/BF01074950.

(43) Yamada, Y.; Furukawa, K.; Sodeyama, K.; Kikuchi, K.; Yaegashi, M.; Tateyama,

Y.; Yamada, A. Unusual Stability of Acetonitrile-Based Superconcentrated

Electrolytes for Fast-Charging Lithium-Ion Batteries. Journal of the American

Chemical Society 2014, 136 (13), 5039–5046. https://doi.org/10.1021/ja412807w.

(44) Yamada, Y.; Wang, J.; Ko, S.; Watanabe, E.; Yamada, A. Advances and Issues in

Developing Salt-Concentrated Battery Electrolytes. Nature Energy 2019.

https://doi.org/10.1038/s41560-019-0336-z.

(45) Sodeyama, K.; Yamada, Y.; Aikawa, K.; Yamada, A.; Tateyama, Y. Sacrificial

Anion Reduction Mechanism for Electrochemical Stability Improvement in Highly

Concentrated Li-Salt Electrolyte. J. Phys. Chem. C 2014, 118 (26), 14091–14097.

https://doi.org/10.1021/jp501178n.

(46) McOwen, D. W.; Seo, D. M.; Borodin, O.; Vatamanu, J.; Boyle, P. D.; Henderson,

W. A. Concentrated Electrolytes: Decrypting Electrolyte Properties and

Reassessing Al Corrosion Mechanisms. Energy Environ. Sci. 2014, 7 (1), 416–426.

https://doi.org/10.1039/C3EE42351D.

(47) Vatamanu, J.; Borodin, O.; Smith, G. D. Molecular Insights into the Potential and

Temperature Dependences of the Differential Capacitance of a Room-Temperature

Ionic Liquid at Graphite Electrodes. J. Am. Chem. Soc. 2010, 132 (42), 14825–

14833. https://doi.org/10.1021/ja104273r.

(48) Jin, J.; Li, H. H.; Wei, J. P.; Bian, X. K.; Zhou, Z.; Yan, J. Li/LiFePO4 Batteries

with Room Temperature Ionic Liquid as Electrolyte. Electrochemistry

Communications 2009, 11 (7), 1500–1503.

https://doi.org/10.1016/j.elecom.2009.05.040.

(49) Yamada, Y.; Chiang, C. H.; Sodeyama, K.; Wang, J.; Tateyama, Y.; Yamada, A.

Corrosion Prevention Mechanism of Aluminum Metal in Superconcentrated

Electrolytes. ChemElectroChem 2015, 2 (11), 1687–1694.

https://doi.org/10.1002/celc.201500235.

References 163

(50) Vatamanu, J.; Borodin, O.; Smith, G. D. Molecular Dynamics Simulation Studies

of the Structure of a Mixed Carbonate/LiPF 6 Electrolyte near Graphite Surface as

a Function of Electrode Potential. J. Phys. Chem. C 2012, 116 (1), 1114–1121.

https://doi.org/10.1021/jp2101539.

(51) Borodin, O.; Ren, X.; Vatamanu, J.; von Wald Cresce, A.; Knap, J.; Xu, K.

Modeling Insight into Battery Electrolyte Electrochemical Stability and Interfacial

Structure. Acc. Chem. Res. 2017, 50 (12), 2886–2894.

https://doi.org/10.1021/acs.accounts.7b00486.

(52) Yamada, Y.; Usui, K.; Chiang, C. H.; Kikuchi, K.; Furukawa, K.; Yamada, A.

General Observation of Lithium Intercalation into Graphite in Ethylene-Carbonate-

Free Superconcentrated Electrolytes. ACS Appl. Mater. Interfaces 2014, 6 (14),

10892–10899. https://doi.org/10.1021/am5001163.

(53) Wang, J.; Yamada, Y.; Sodeyama, K.; Watanabe, E.; Takada, K.; Tateyama, Y.;

Yamada, A. Fire-Extinguishing Organic Electrolytes for Safe Batteries. Nature

Energy 2018, 3 (1), 22–29. https://doi.org/10.1038/s41560-017-0033-8.

(54) Yamada, Y.; Yamada, A. Superconcentrated Electrolytes to Create New Interfacial

Chemistry in Non-Aqueous and Aqueous Rechargeable Batteries. Chemistry

Letters 2017, 46 (8), 1056–1064. https://doi.org/10.1246/cl.170284.

(55) Qian, J.; Henderson, W. A.; Xu, W.; Bhattacharya, P.; Engelhard, M.; Borodin, O.;

Zhang, J.-G. High Rate and Stable Cycling of Lithium Metal Anode. Nat Commun

2015, 6 (1), 6362. https://doi.org/10.1038/ncomms7362.

(56) Alvarado, J.; Schroeder, M. A.; Zhang, M.; Borodin, O.; Gobrogge, E.; Olguin, M.;

Ding, M. S.; Gobet, M.; Greenbaum, S.; Meng, Y. S.; Xu, K. A Carbonate-Free,

Sulfone-Based Electrolyte for High-Voltage Li-Ion Batteries. Materials Today

2018, 21 (4), 341–353. https://doi.org/10.1016/j.mattod.2018.02.005.

(57) Wang, J.; Yamada, Y.; Sodeyama, K.; Chiang, C. H.; Tateyama, Y.; Yamada, A.

Superconcentrated Electrolytes for a High-Voltage Lithium-Ion Battery. Nature

Communications 2016, 7, 12032. https://doi.org/10.1038/ncomms12032.

(58) Yamada, Y.; Usui, K.; Sodeyama, K.; Ko, S.; Tateyama, Y.; Yamada, A. Hydrate-

Melt Electrolytes for High-Energy-Density Aqueous Batteries. Nature Energy

2016, 1 (10). https://doi.org/10.1038/nenergy.2016.129.

(59) McKinnon, W. R.; Dahn, J. R. How to Reduce the Cointercalation of Propylene

Carbonate in Li x ZrS2 and Other Layered Compounds. J. Electrochem. Soc. 1985,

132 (2), 364–366. https://doi.org/10.1149/1.2113839.

(60) Jeong, S.-K.; Inaba, M.; Iriyama, Y.; Abe, T.; Ogumi, Z. Electrochemical

Intercalation of Lithium Ion within Graphite from Propylene Carbonate Solutions.

Electrochem. Solid-State Lett. 2003, 6 (1), A13.

https://doi.org/10.1149/1.1526781.

164 References

(61) Yamada, Y.; Takazawa, Y.; Miyazaki, K.; Abe, T. Electrochemical Lithium

Intercalation into Graphite in Dimethyl Sulfoxide-Based Electrolytes: Effect of

Solvation Structure of Lithium Ion. J. Phys. Chem. C 2010, 114 (26), 11680–

11685. https://doi.org/10.1021/jp1037427.

(62) Yamada, Y.; Yaegashi, M.; Abe, T.; Yamada, A. A Superconcentrated Ether

Electrolyte for Fast-Charging Li-Ion Batteries. Chemical Communications 2013,

49 (95), 11194. https://doi.org/10.1039/c3cc46665e.

(63) Petibon, R.; Aiken, C. P.; Ma, L.; Xiong, D.; Dahn, J. R. The Use of Ethyl Acetate

as a Sole Solvent in Highly Concentrated Electrolyte for Li-Ion Batteries.

Electrochimica Acta 2015, 154, 287–293.

https://doi.org/10.1016/j.electacta.2014.12.093.

(64) Lu, D.; Tao, J.; Yan, P.; Henderson, W. A.; Li, Q.; Shao, Y.; Helm, M. L.; Borodin,

O.; Graff, G. L.; Polzin, B.; Wang, C.-M.; Engelhard, M.; Zhang, J.-G.; De Yoreo,

J. J.; Liu, J.; Xiao, J. Formation of Reversible Solid Electrolyte Interface on

Graphite Surface from Concentrated Electrolytes. Nano Lett. 2017, 17 (3), 1602–

1609. https://doi.org/10.1021/acs.nanolett.6b04766.

(65) Chen, S.; Zheng, J.; Yu, L.; Ren, X.; Engelhard, M. H.; Niu, C.; Lee, H.; Xu, W.;

Xiao, J.; Liu, J.; Zhang, J.-G. High-Efficiency Lithium Metal Batteries with Fire-

Retardant Electrolytes. Joule 2018, 2 (8), 1548–1558.

https://doi.org/10.1016/j.joule.2018.05.002.

(66) Angell, C. A.; Liu, C.; Sanchez, E. Rubbery Solid Electrolytes with Dominant

Cationic Transport and High Ambient Conductivity. Nature 1993, 362, 137–139.

(67) Tamura, T.; Yoshida, K.; Hachida, T.; Tsuchiya, M.; Nakamura, M.; Kazue, Y.;

Tachikawa, N.; Dokko, K.; Watanabe, M. Physicochemical Properties of Glyme–

Li Salt Complexes as a New Family of Room-Temperature Ionic Liquids. Chem.

Lett. 2010, 39 (7), 753–755. https://doi.org/10.1246/cl.2010.753.

(68) Tamura, T.; Hachida, T.; Yoshida, K.; Tachikawa, N.; Dokko, K.; Watanabe, M.

New Glyme–Cyclic Imide Lithium Salt Complexes as Thermally Stable

Electrolytes for Lithium Batteries. Journal of Power Sources 2010, 195 (18),

6095–6100. https://doi.org/10.1016/j.jpowsour.2009.11.061.

(69) Yoshida, K.; Nakamura, M.; Kazue, Y.; Tachikawa, N.; Tsuzuki, S.; Seki, S.;

Dokko, K.; Watanabe, M. Oxidative-Stability Enhancement and Charge Transport

Mechanism in Glyme–Lithium Salt Equimolar Complexes. J. Am. Chem. Soc.

2011, 133 (33), 13121–13129. https://doi.org/10.1021/ja203983r.

(70) Doi, T.; Masuhara, R.; Hashinokuchi, M.; Shimizu, Y.; Inaba, M. Concentrated

LiPF6/PC Electrolyte Solutions for 5-V LiNi0.5Mn1.5O4 Positive Electrode in

Lithium-Ion Batteries. Electrochimica Acta 2016, 209, 219–224.

https://doi.org/10.1016/j.electacta.2016.05.062.

References 165

(71) Suo, L.; Xue, W.; Gobet, M.; Greenbaum, S. G.; Wang, C.; Chen, Y.; Yang, W.;

Li, Y.; Li, J. Fluorine-Donating Electrolytes Enable Highly Reversible 5-V-Class

Li Metal Batteries. Proc Natl Acad Sci USA 2018, 115 (6), 1156–1161.

https://doi.org/10.1073/pnas.1712895115.

(72) Matsumoto, K.; Inoue, K.; Nakahara, K.; Yuge, R.; Noguchi, T.; Utsugi, K.

Suppression of Aluminum Corrosion by Using High Concentration LiTFSI

Electrolyte. Journal of Power Sources 2013, 231, 234–238.

https://doi.org/10.1016/j.jpowsour.2012.12.028.

(73) Shiga, T.; Okuda, C.; Kato, Y.; Kondo, H. Highly Concentrated Electrolytes

Containing a Phosphoric Acid Ester Amide with Self-Extinguishing Properties for

Use in Lithium Batteries. J. Phys. Chem. C 2018, 122 (18), 9738–9745.

https://doi.org/10.1021/acs.jpcc.7b12461.

(74) Hess, S.; Wohlfahrt-Mehrens, M.; Wachtler, M. Flammability of Li-Ion Battery

Electrolytes: Flash Point and Self-Extinguishing Time Measurements. J.

Electrochem. Soc. 2015, 162 (2), A3084–A3097.

https://doi.org/10.1149/2.0121502jes.

(75) Nakayama, N.; Nozawa, T.; Iriyama, Y.; Abe, T.; Ogumi, Z.; Kikuchi, K.

Interfacial Lithium-Ion Transfer at the LiMn2O4 Thin Film Electrode/Aqueous

Solution Interface. Journal of Power Sources 2007, 174 (2), 695–700.

https://doi.org/10.1016/j.jpowsour.2007.06.113.

(76) Sauvage, F.; Laffont, L.; Tarascon, J.-M.; Baudrin, E. Factors Affecting the

Electrochemical Reactivity vs. Lithium of Carbon-Free LiFePO4 Thin Films.

Journal of Power Sources 2008, 175 (1), 495–501.

https://doi.org/10.1016/j.jpowsour.2007.09.085.

(77) Tang, W.; Tian, S.; Liu, L. L.; Li, L.; Zhang, H. P.; Yue, Y. B.; Bai, Y.; Wu, Y. P.;

Zhu, K. Nanochain LiMn2O4 as Ultra-Fast Cathode Material for Aqueous

Rechargeable Lithium Batteries. Electrochemistry Communications 2011, 13 (2),

205–208. https://doi.org/10.1016/j.elecom.2010.12.015.

(78) Qu, Q.; Fu, L.; Zhan, X.; Samuelis, D.; Maier, J.; Li, L.; Tian, S.; Li, Z.; Wu, Y.

Porous LiMn2O4 as Cathode Material with High Power and Excellent Cycling for

Aqueous Rechargeable Lithium Batteries. Energy Environ. Sci. 2011, 4 (10), 3985.

https://doi.org/10.1039/c0ee00673d.

(79) Alias, N.; Mohamad, A. A. Advances of Aqueous Rechargeable Lithium-Ion

Battery: A Review. Journal of Power Sources 2015, 274, 237–251.

https://doi.org/10.1016/j.jpowsour.2014.10.009.

(80) Li, W.; Dahn, J. R.; Wainwright, D. S. Rechargeable Lithium Batteries with

Aqueous Electrolytes. Science 1994, 264 (5162), 1115–1118.

https://doi.org/10.1126/science.264.5162.1115.

166 References

(81) Tian, L.; Yuan, A. Electrochemical Performance of Nanostructured Spinel

LiMn2O4 in Different Aqueous Electrolytes. Journal of Power Sources 2009, 192

(2), 693–697. https://doi.org/10.1016/j.jpowsour.2009.03.002.

(82) Yadegari, H.; Jabbari, A.; Heli, H. An Aqueous Rechargeable Lithium-Ion Battery

Based on LiCoO2 Nanoparticles Cathode and LiV3O8 Nanosheets Anode. J Solid

State Electrochem 2012, 16 (1), 227–234. https://doi.org/10.1007/s10008-011-

1315-x.

(83) Ruffo, R.; Wessells, C.; Huggins, R. A.; Cui, Y. Electrochemical Behavior of

LiCoO2 as Aqueous Lithium-Ion Battery Electrodes. Electrochemistry

Communications 2009, 11 (2), 247–249.

https://doi.org/10.1016/j.elecom.2008.11.015.

(84) Ruffo, R.; La Mantia, F.; Wessells, C.; Huggins, R. A.; Cui, Y. Electrochemical

Characterization of LiCoO2 as Rechargeable Electrode in Aqueous LiNO3

Electrolyte. Solid State Ionics 2011, 192 (1), 289–292.

https://doi.org/10.1016/j.ssi.2010.05.043.

(85) Zhao, M. Excellent Rate Capabilities of (LiFePO4/C)//LiV3O8 in an Optimized

Aqueous Solution Electrolyte. Journal of Power Sources 2013, 6.

(86) Vujković, M.; Stojković, I.; Cvjetićanin, N.; Mentus, S. Gel-Combustion Synthesis

of LiFePO4/C Composite with Improved Capacity Retention in Aerated Aqueous

Electrolyte Solution. Electrochimica Acta 2013, 92, 248–256.

https://doi.org/10.1016/j.electacta.2013.01.030.

(87) Zhao, M.; Huang, G.; Zhang, B.; Wang, F.; Song, X. Characteristics and

Electrochemical Performance of LiFe0.5Mn0.5PO4/C Used as Cathode for Aqueous

Rechargeable Lithium Battery. Journal of Power Sources 2012, 211, 202–207.

https://doi.org/10.1016/j.jpowsour.2012.03.049.

(88) Wang, H.; Zeng, Y.; Huang, K.; Liu, S.; Chen, L. Improvement of Cycle

Performance of Lithium Ion Cell LiMn2O4/LixV2O5 with Aqueous Solution

Electrolyte by Polypyrrole Coating on Anode. Electrochimica Acta 2007, 52 (15),

5102–5107. https://doi.org/10.1016/j.electacta.2007.02.004.

(89) Wessells, C.; Huggins, R. A.; Cui, Y. Recent Results on Aqueous Electrolyte Cells.

Journal of Power Sources 2011, 196 (5), 2884–2888.

https://doi.org/10.1016/j.jpowsour.2010.10.098.

(90) Tang, W.; Liu, L. L.; Tian, S.; Li, L.; Li, L. L.; Yue, Y. B.; Bai, Y.; Wu, Y. P.;

Zhu, K.; Holze, R. LiMn2O4 Nanorods as a Super-Fast Cathode Material for

Aqueous Rechargeable Lithium Batteries. Electrochemistry Communications

2011, 13 (11), 1159–1162. https://doi.org/10.1016/j.elecom.2011.09.008.

(91) Luo, J.-Y.; Xia, Y.-Y. Aqueous Lithium-Ion Battery LiTi2(PO4)3/LiMn2O4 with

High Power and Energy Densities as Well as Superior Cycling Stability**. Adv.

Funct. Mater. 2007, 17 (18), 3877–3884. https://doi.org/10.1002/adfm.200700638.

References 167

(92) Ramanujapuram, A.; Gordon, D.; Magasinski, A.; Ward, B.; Nitta, N.; Huang, C.;

Yushin, G. Degradation and Stabilization of Lithium Cobalt Oxide in Aqueous

Electrolytes. Energy Environ. Sci. 2016, 9 (5), 1841–1848.

https://doi.org/10.1039/C6EE00093B.

(93) Mi, C. H.; Zhang, X. G.; Li, H. L. Electrochemical Behaviors of Solid LiFePO4

and Li0.99Nb0.01FePO4 in Li2SO4 Aqueous Electrolyte. Journal of Electroanalytical

Chemistry 2007, 602 (2), 245–254.

https://doi.org/10.1016/j.jelechem.2007.01.007.

(94) Mahesh, K. C.; Suresh, G. S.; Venkatesha, T. V. Electrochemical Behavior of

Li[Li0.2Co0.3Mn0.5]O2 as Cathode Material in Li2SO4 Aqueous Electrolyte. J Solid

State Electrochem 2012, 16 (11), 3559–3571. https://doi.org/10.1007/s10008-012-

1787-3.

(95) Zhou, D.; Liu, S.; Wang, H.; Yan, G. Na2V6O16·0.14H2O Nanowires as a Novel

Anode Material for Aqueous Rechargeable Lithium Battery with Good Cycling

Performance. Journal of Power Sources 2013, 227, 111–117.

https://doi.org/10.1016/j.jpowsour.2012.11.022.

(96) Liu, X.-H.; Saito, T.; Doi, T.; Okada, S.; Yamaki, J. Electrochemical Properties of

Rechargeable Aqueous Lithium Ion Batteries with an Olivine-Type Cathode and a

Nasicon-Type Anode. Journal of Power Sources 2009, 189 (1), 706–710.

https://doi.org/10.1016/j.jpowsour.2008.08.050.

(97) Manjunatha, H.; Suresh, G. S.; Venkatesha, T. V. Electrode Materials for Aqueous

Rechargeable Lithium Batteries. J Solid State Electrochem 2011, 15 (3), 431–445.

https://doi.org/10.1007/s10008-010-1117-6.

(98) Degoulange, D.; Dubouis, N.; Grimaud, A. Towards the Understanding of Water-

in-Salt Electrolytes: Individual Ion Activities and Liquid Junction Potentials in

Highly Concentrated Aqueous Solutions. Physical Chemistry 2021, 33.

(99) Zhao, M. Electrochemical Performance of High Specific Capacity of Lithium-Ion

Cell LiV3O8//LiMn2O4 with LiNO3 Aqueous Solution Electrolyte. Electrochimica

Acta 2011, 4.

(100) Wang, G. J.; Zhao, N. H.; Yang, L. C.; Wu, Y. P.; Wu, H. Q.; Holze, R.

Characteristics of an Aqueous Rechargeable Lithium Battery (ARLB).

Electrochimica Acta 2007, 52 (15), 4911–4915.

https://doi.org/10.1016/j.electacta.2007.01.051.

(101) Li, W.; McKinnon, W. R.; Dahn, J. R. Lithium Intercalation from Aqueous

Solutions. J. Electrochem. Soc. 1994, 141 (9), 7.

(102) Li, W.; Dahn, J. R. Lithium‐Ion Cells with Aqueous Electrolytes. J. Electrochem.

Soc. 1995, 142 (6), 1742–1746. https://doi.org/10.1149/1.2044187.

168 References

(103) Wang, P.; Yang, H.; Yang, H. Electrochemical Behavior of Li-Mn Spine1

Electrode Material in Aqueous Solution. Journal of Power Sources 1996, 63, 275–

278.

(104) Yu, D. Y. W. Impurities in LiFePO4 and Their Influence on Material

Characteristics. Meet. Abstr. 2007. https://doi.org/10.1149/MA2007-02/10/646.

(105) Manickam, M.; Singh, P.; Thurgate, S.; Prince, K. Redox Behavior and Surface

Characterization of LiFePO4 in Lithium Hydroxide Electrolyte. Journal of Power

Sources 2006, 158 (1), 646–649. https://doi.org/10.1016/j.jpowsour.2005.08.059.

(106) He, P.; Wang, Y.; Zhou, H. The Effect of Alkalinity and Temperature on the

Performance of Lithium-Air Fuel Cell with Hybrid Electrolytes. Journal of Power

Sources 2011, 196 (13), 5611–5616.

https://doi.org/10.1016/j.jpowsour.2011.02.071.

(107) Luo, J.-Y.; Cui, W.-J.; He, P.; Xia, Y.-Y. Raising the Cycling Stability of Aqueous

Lithium-Ion Batteries by Eliminating Oxygen in the Electrolyte. Nature Chem

2010, 2 (9), 760–765. https://doi.org/10.1038/nchem.763.

(108) Manthiram, A.; Choi, J. Chemical and Structural Instabilities of Lithium Ion

Battery Cathodes. Journal of Power Sources 2006, 159 (1), 249–253.

https://doi.org/10.1016/j.jpowsour.2006.04.028.

(109) Bin, D.; Wen, Y.; Wang, Y.; Xia, Y. The Development in Aqueous Lithium-Ion

Batteries. Journal of Energy Chemistry 2018, 27 (6), 1521–1535.

https://doi.org/10.1016/j.jechem.2018.06.004.

(110) Wang, Y.; Luo, J.; Wang, C.; Xia, Y. Hybrid Aqueous Energy Storage Cells Using

Activated Carbon and Lithium-Ion Intercalated Compounds. J. Electrochem. Soc.

2006, 153 (8), A1425. https://doi.org/10.1149/1.2203772.

(111) Suo, L.; Borodin, O.; Gao, T.; Olguin, M.; Ho, J.; Fan, X.; Luo, C.; Wang, C.; Xu,

K. “Water-in-Salt” Electrolyte Enables High-Voltage Aqueous Lithium-Ion

Chemistries. Science 2015, 350 (6263), 938–943.

https://doi.org/10.1126/science.aab1595.

(112) Wang, H.; Huang, K.; Zeng, Y.; Yang, S.; Chen, L. Electrochemical Properties of

TiP2O7 and LiTi2(PO4)3 as Anode Material for Lithium Ion Battery with Aqueous

Solution Electrolyte. Electrochimica Acta 2007, 52 (9), 3280–3285.

https://doi.org/10.1016/j.electacta.2006.10.010.

(113) Wojciechowski, J.; Kolanowski, Ł.; Bund, A.; Lota, G. The Influence of Current

Collector Corrosion on the Performance of Electrochemical Capacitors. Journal of

Power Sources 2017, 368, 18–29. https://doi.org/10.1016/j.jpowsour.2017.09.069.

(114) Hou, Y.; Wang, X.; Zhu, Y.; Hu, C.; Chang, Z.; Wu, Y.; Holze, R. Macroporous

LiFePO4 as a Cathode for an Aqueous Rechargeable Lithium Battery of High

References 169

Energy Density. J. Mater. Chem. A 2013, 1 (46), 14713.

https://doi.org/10.1039/c3ta13472e.

(115) He, K.; Zu, C.; Wang, Y.; Han, B.; Yin, X.; Zhao, H.; Liu, Y.; Chen, J. Stability of

Lithium Ion Conductor NASICON Structure Glass Ceramic in Acid and Alkaline

Aqueous Solution. Solid State Ionics 2014, 254, 78–81.

https://doi.org/10.1016/j.ssi.2013.11.011.

(116) Imanishi, N.; Matsui, M.; Takeda, Y.; Yamamoto, O. Lithium Ion Conducting

Solid Electrolytes for Aqueous Lithium-Air Batteries. Electrochemistry 2014, 82

(11), 938–945. https://doi.org/10.5796/electrochemistry.82.938.

(117) Wang, X.; Qu, Q.; Hou, Y.; Wang, F.; Wu, Y. An Aqueous Rechargeable Lithium

Battery of High Energy Density Based on Coated Li Metal and LiCoO2. Chem.

Commun. 2013, 49 (55), 6179. https://doi.org/10.1039/c3cc42676a.

(118) Chang, Z.; Li, C.; Wang, Y.; Chen, B.; Fu, L.; Zhu, Y.; Zhang, L.; Wu, Y.; Huang,

W. A Lithium Ion Battery Using an Aqueous Electrolyte Solution. Sci Rep 2016,

6 (1), 28421. https://doi.org/10.1038/srep28421.

(119) Wang, X.; Hou, Y.; Zhu, Y.; Wu, Y.; Holze, R. An Aqueous Rechargeable Lithium

Battery Using Coated Li Metal as Anode. SCIENTIFIC REPORTS 5.

(120) Alloin, F.; Crepel, L.; Cointeaux, L.; Leprêtre, J.-C.; Fusalba, F.; Martinet, S. The

Interest of Diazonium Chemistry for Aqueous Lithium-Ion Battery. J.

Electrochem. Soc. 2013, 160 (5), A3171–A3178.

https://doi.org/10.1149/2.026305jes.

(121) Crepel, L.; Alloin, F.; Lepretre, J.-C.; Martinet, S.; Ii, H. (54) ARYL DAZONUMI

SALT AND USE IN AN ELECTROLYTIC SOLUTION OF AN

ELECTROCHEMICAL GENERATOR. 15.

(122) Suo, L.; Borodin, O.; Sun, W.; Fan, X.; Yang, C.; Wang, F.; Gao, T.; Ma, Z.;

Schroeder, M.; von Cresce, A.; Russell, S. M.; Armand, M.; Angell, A.; Xu, K.;

Wang, C. Advanced High-Voltage Aqueous Lithium-Ion Battery Enabled by

“Water-in-Bisalt” Electrolyte. Angewandte Chemie International Edition 2016, 55

(25), 7136–7141. https://doi.org/10.1002/anie.201602397.

(123) Han, S. Anionic Effects on the Structure and Dynamics of Water in

Superconcentrated Aqueous Electrolytes. RSC Adv. 2019, 9 (2), 609–619.

https://doi.org/10.1039/C8RA09589B.

(124) Lux, S. F.; Terborg, L.; Hachmöller, O.; Placke, T.; Meyer, H.-W.; Passerini, S.;

Winter, M.; Nowak, S. LiTFSI Stability in Water and Its Possible Use in Aqueous

Lithium-Ion Batteries: PH Dependency, Electrochemical Window and

Temperature Stability. Journal of The Electrochemical Society 2013, 160 (10),

A1694–A1700. https://doi.org/10.1149/2.039310jes.

170 References

(125) Reber, D.; Figi, R.; Kühnel, R.-S.; Battaglia, C. Stability of Aqueous Electrolytes

Based on LiFSI and NaFSI. Electrochimica Acta 2019, 321, 134644.

https://doi.org/10.1016/j.electacta.2019.134644.

(126) Ding, M. S.; Xu, K. Phase Diagram, Conductivity, and Glass Transition of LiTFSI–

H 2 O Binary Electrolytes. J. Phys. Chem. C 2018, 122 (29), 16624–16629.

https://doi.org/10.1021/acs.jpcc.8b05193.

(127) Dubouis, N.; France-Lanord, A.; Brige, A.; Salanne, M.; Grimaud, A. Anion

Specific Effects Drive the Formation of Li-Salt Based Aqueous Biphasic Systems.

J. Phys. Chem. B 2021, acs.jpcb.1c01750.

https://doi.org/10.1021/acs.jpcb.1c01750.

(128) Zhang, Y.; Lewis, N. H. C.; Mars, J.; Wan, G.; Weadock, N. J.; Takacs, C. J.;

Lukatskaya, M. R.; Steinrück, H.-G.; Toney, M. F.; Tokmakoff, A.; Maginn, E. J.

Water-in-Salt LiTFSI Aqueous Electrolytes. 1. Liquid Structure from Combined

Molecular Dynamics Simulation and Experimental Studies. J. Phys. Chem. B 2021,

125 (17), 4501–4513. https://doi.org/10.1021/acs.jpcb.1c02189.

(129) Lewis, N. H. C.; Zhang, Y.; Dereka, B.; Carino, E. V.; Maginn, E. J.; Tokmakoff,

A. Signatures of Ion Pairing and Aggregation in the Vibrational Spectroscopy of

Super-Concentrated Aqueous Lithium Bistriflimide Solutions. J. Phys. Chem. C

2020, 124 (6), 3470–3481. https://doi.org/10.1021/acs.jpcc.9b10477.

(130) Tsurumura, T.; Hashimoto, Y.; Morita, M.; Umebayashi, Y.; Fujii, K. Anion

Coordination Characteristics of Ion-Pair Complexes in Highly Concentrated

Aqueous Lithium Bis(Trifluoromethanesulfonyl)Amide Electrolytes. Anal. Sci.

2019, 35 (3), 289–294. https://doi.org/10.2116/analsci.18P407.

(131) Liu, X.; Yu, Z.; Sarnello, E.; Qian, K.; Seifert, S.; Winans, R. E.; Cheng, L.; Li, T.

Microscopic Understanding of the Ionic Networks of “Water-in-Salt” Electrolytes.

Energy Material Advances 2021, 2021, 1–9.

https://doi.org/10.34133/2021/7368420.

(132) Watanabe, H.; Arai, N.; Nozaki, E.; Han, J.; Fujii, K.; Ikeda, K.; Otomo, T.; Ueno,

K.; Dokko, K.; Watanabe, M.; Kameda, Y.; Umebayashi, Y. Local Structure of Li + in Superconcentrated Aqueous LiTFSA Solutions. J. Phys. Chem. B 2021, 125

(27), 7477–7484. https://doi.org/10.1021/acs.jpcb.1c04693.

(133) González, M. A.; Borodin, O.; Kofu, M.; Shibata, K.; Yamada, T.; Yamamuro, O.;

Xu, K.; Price, D. L.; Saboungi, M.-L. Nanoscale Relaxation in “Water-in-Salt” and

“Water-in-Bisalt” Electrolytes. J. Phys. Chem. Lett. 2020, 11 (17), 7279–7284.

https://doi.org/10.1021/acs.jpclett.0c01765.

(134) Dubouis, N.; Lemaire, P.; Mirvaux, B.; Salager, E.; Deschamps, M.; Grimaud, A.

The Role of the Hydrogen Evolution Reaction in the Solid–Electrolyte Interphase

Formation Mechanism for “ Water-in-Salt ” Electrolytes. Energy & Environmental

Science 2018, 11 (12), 3491–3499. https://doi.org/10.1039/C8EE02456A.

References 171

(135) Borodin, O.; Suo, L.; Gobet, M.; Ren, X.; Wang, F.; Faraone, A.; Peng, J.; Olguin,

M.; Schroeder, M.; Ding, M. S.; Gobrogge, E.; von Wald Cresce, A.; Munoz, S.;

Dura, J. A.; Greenbaum, S.; Wang, C.; Xu, K. Liquid Structure with Nano-

Heterogeneity Promotes Cationic Transport in Concentrated Electrolytes. ACS

Nano 2017, 11 (10), 10462–10471. https://doi.org/10.1021/acsnano.7b05664.

(136) Yu, Z.; Curtiss, L. A.; Winans, R. E.; Zhang, Y.; Li, T.; Cheng, L. Asymmetric

Composition of Ionic Aggregates and the Origin of High Correlated Transference

Number in Water-in-Salt Electrolytes. J. Phys. Chem. Lett. 2020, 11 (4), 1276–

1281. https://doi.org/10.1021/acs.jpclett.9b03495.

(137) Lim, J.; Park, K.; Lee, H.; Kim, J.; Kwak, K.; Cho, M. Nanometric Water Channels

in Water-in-Salt Lithium Ion Battery Electrolyte. J. Am. Chem. Soc. 2018, 140 (46),

15661–15667. https://doi.org/10.1021/jacs.8b07696.

(138) Tan, P.; Yue, J.; Yu, Y.; Liu, B.; Liu, T.; Zheng, L.; He, L.; Zhang, X.; Suo, L.;

Hong, L. Solid-Like Nano-Anion Cluster Constructs a Free Lithium-Ion-

Conducting Superfluid Framework in a Water-in-Salt Electrolyte. J. Phys. Chem.

C 2021, 125 (22), 11838–11847. https://doi.org/10.1021/acs.jpcc.1c01663.

(139) Zhang, M.; Hao, H.; Zhou, D.; Duan, Y.; Wang, Y.; Bian, H. Understanding the

Microscopic Structure of a “Water-in-Salt” Lithium Ion Battery Electrolyte Probed

with Ultrafast IR Spectroscopy. J. Phys. Chem. C 2020, 124 (16), 8594–8604.

https://doi.org/10.1021/acs.jpcc.0c00937.

(140) McEldrew, M.; Goodwin, Z. A. H.; Bi, S.; Kornyshev, A. A.; Bazant, M. Z. Ion

Clusters and Networks in “Water-in-Salt Electrolytes.” arXiv:2103.04782 [cond-

mat, physics:physics] 2021.

(141) Li, Z.; Bouchal, R.; Mendez-Morales, T.; Rollet, A.-L.; Rizzi, C.; Le Vot, S.;

Favier, F.; Rotenberg, B.; Borodin, O.; Fontaine, O.; Salanne, M. Transport

Properties of Li-TFSI Water-in-Salt Electrolytes. J. Phys. Chem. B 2019, 123 (49),

10514–10521. https://doi.org/10.1021/acs.jpcb.9b08961.

(142) Amiri, M.; Bélanger, D. Physicochemical and Electrochemical Properties of

Water‐in‐Salt Electrolytes. ChemSusChem 2021, 14 (12), 2487–2500.

https://doi.org/10.1002/cssc.202100550.

(143) Zheng, J.; Tan, G.; Shan, P.; Liu, T.; Hu, J.; Feng, Y.; Yang, L.; Zhang, M.; Chen,

Z.; Lin, Y.; Lu, J.; Neuefeind, J. C.; Ren, Y.; Amine, K.; Wang, L.-W.; Xu, K.;

Pan, F. Understanding Thermodynamic and Kinetic Contributions in Expanding

the Stability Window of Aqueous Electrolytes. Chem 2018.

https://doi.org/10.1016/j.chempr.2018.09.004.

(144) Dubouis, N.; Serva, A.; Salager, E.; Deschamps, M.; Salanne, M.; Grimaud, A.

The Fate of Water at the Electrochemical Interfaces: Electrochemical Behavior of

Free Water Versus Coordinating Water. J. Phys. Chem. Lett. 2018, 6.

172 References

(145) Han, M.; Zhang, R.; Gewirth, A. A.; Espinosa-Marzal, R. M. Nanoheterogeneity

of LiTFSI Solutions Transitions Close to a Surface and with Concentration. Nano

Lett. 2021, 21 (5), 2304–2309. https://doi.org/10.1021/acs.nanolett.1c00167.

(146) Suo, L.; Oh, D.; Lin, Y.; Zhuo, Z.; Borodin, O.; Gao, T.; Wang, F.; Kushima, A.;

Wang, Z.; Kim, H.-C.; Qi, Y.; Yang, W.; Pan, F.; Li, J.; Xu, K.; Wang, C. How

Solid-Electrolyte Interphase Forms in Aqueous Electrolytes. Journal of the

American Chemical Society 2017, 139 (51), 18670–18680.

https://doi.org/10.1021/jacs.7b10688.

(147) Steinrück, H.; Cao, C.; Lukatskaya, M. R.; Takacs, C. J.; Wan, G.; Mackanic, D.

G.; Tsao, Y.; Zhao, J.; Helms, B. A.; Xu, K.; Borodin, O.; Wishart, J. F.; Toney,

M. F. Interfacial Speciation Determines Interfacial Chemistry: X‐ray‐Induced

Lithium Fluoride Formation from Water‐in‐salt Electrolytes on Solid Surfaces.

Angew. Chem. Int. Ed. 2020, anie.202007745.

https://doi.org/10.1002/anie.202007745.

(148) Ko, S.; Yamada, Y.; Yamada, A. Formation of a Solid Electrolyte Interphase in

Hydrate-Melt Electrolytes. ACS Appl. Mater. Interfaces 2019, acsami.9b13662.

https://doi.org/10.1021/acsami.9b13662.

(149) Bouchal, R.; Li, Z.; Bongu, C.; Favier, F.; Freunberger, S. A.; Salanne, M.;

Fontaine, O. Competitive Salt Precipitation/Dissolution during Free-Water

Reduction in Water-in-Salt Electrolyte. Angewandte Chemie International Edition

2020, 59, 6.

(150) Vatamanu, J.; Borodin, O. Ramifications of Water-in-Salt Interfacial Structure at

Charged Electrodes for Electrolyte Electrochemical Stability. The Journal of

Physical Chemistry Letters 2017, 8 (18), 4362–4367.

https://doi.org/10.1021/acs.jpclett.7b01879.

(151) Zhang, R.; Han, M.; Ta, K.; Madsen, K. E.; Chen, X.; Zhang, X.; Espinosa-Marzal,

R. M.; Gewirth, A. A. Potential-Dependent Layering in the Electrochemical

Double Layer of Water-in-Salt Electrolytes. ACS Appl. Energy Mater. 2020, 3 (8),

8086–8094. https://doi.org/10.1021/acsaem.0c01534.

(152) Coustan, L.; Shul, G.; Bélanger, D. Electrochemical Behavior of Platinum, Gold

and Glassy Carbon Electrodes in Water-in-Salt Electrolyte. Electrochemistry

Communications 2017, 77, 89–92. https://doi.org/10.1016/j.elecom.2017.03.001.

(153) Kühnel, R.-S.; Reber, D.; Battaglia, C. Perspective—Electrochemical Stability of

Water-in-Salt Electrolytes. J. Electrochem. Soc. 2020, 167 (7), 070544.

https://doi.org/10.1149/1945-7111/ab7c6f.

(154) Takenaka, N.; Inagaki, T.; Shimada, T.; Yamada, Y.; Nagaoka, M.; Yamada, A.

Theoretical Analysis of Electrode-Dependent Interfacial Structures on Hydrate-

Melt Electrolytes. J Chem. Phys. 2020, 152 (12), 124706.

https://doi.org/10.1063/5.0003196.

References 173

(155) McEldrew, M.; Goodwin, Z. A. H.; Kornyshev, A. A.; Bazant, M. Z. Theory of the

Double Layer in Water-in-Salt Electrolytes. The Journal of Physical Chemistry

Letters 2018, 9 (19), 5840–5846. https://doi.org/10.1021/acs.jpclett.8b02543.

(156) Liu, S.; Liu, D.; Wang, S.; Cai, X.; Qian, K.; Kang, F.; Li, B. Understanding the

Cathode Electrolyte Interface Formation in Aqueous Electrolyte by Scanning

Electrochemical Microscopy. J. Mater. Chem. A 2019, 7 (21), 12993–12996.

https://doi.org/10.1039/C9TA03199E.

(157) Sun, W.; Suo, L.; Wang, F.; Eidson, N.; Yang, C.; Han, F.; Ma, Z.; Gao, T.; Zhu,

M.; Wang, C. “Water-in-Salt” Electrolyte Enabled LiMn2O4/TiS2 Lithium-Ion

Batteries. Electrochemistry Communications 2017, 82, 71–74.

https://doi.org/10.1016/j.elecom.2017.07.016.

(158) Yue, J.; Lin, L.; Jiang, L.; Zhang, Q.; Tong, Y.; Suo, L.; Hu, Y.; Li, H.; Huang, X.;

Chen, L. Interface Concentrated‐Confinement Suppressing Cathode Dissolution in

Water‐in‐Salt Electrolyte. Adv. Energy Mater. 2020, 10 (36), 2000665.

https://doi.org/10.1002/aenm.202000665.

(159) Wang, F.; Lin, Y.; Suo, L.; Fan, X.; Gao, T.; Yang, C.; Han, F.; Qi, Y.; Xu, K.;

Wang, C. Stabilizing High Voltage LiCoO2 Cathode in Aqueous Electrolyte with

Interphase-Forming Additive. Energy & Environmental Science 2016, 9 (12),

3666–3673. https://doi.org/10.1039/C6EE02604D.

(160) Kühnel, R.-S.; Reber, D.; Remhof, A.; Figi, R.; Bleiner, D.; Battaglia, C. “Water-

in-Salt” Electrolytes Enable the Use of Cost-Effective Aluminum Current

Collectors for Aqueous High-Voltage Batteries. Chemical Communications 2016,

52 (68), 10435–10438. https://doi.org/10.1039/C6CC03969C.

(161) Han, J.; Zhang, H.; Varzi, A.; Passerini, S. Fluorine-Free Water-in-Salt Electrolyte

for Green and Low-Cost Aqueous Sodium-Ion Batteries. ChemSusChem 2018, 11

(21), 3704–3707. https://doi.org/10.1002/cssc.201801930.

(162) Jiang, L.; Liu, L.; Yue, J.; Zhang, Q.; Zhou, A.; Borodin, O.; Suo, L.; Li, H.; Chen,

L.; Xu, K.; Hu, Y. High‐Voltage Aqueous Na‐Ion Battery Enabled by Inert‐Cation‐

Assisted Water‐in‐Salt Electrolyte. Adv. Mater. 2020, 32 (2), 1904427.

https://doi.org/10.1002/adma.201904427.

(163) Reber, D.; Kühnel, R.-S.; Battaglia, C. Suppressing Crystallization of Water-in-

Salt Electrolytes by Asymmetric Anions Enables Low-Temperature Operation of

High-Voltage Aqueous Batteries. ACS Materials Lett. 2019, 1 (1), 44–51.

https://doi.org/10.1021/acsmaterialslett.9b00043.

(164) Suo, L.; Borodin, O.; Wang, Y.; Rong, X.; Sun, W.; Fan, X.; Xu, S.; Schroeder, M.

A.; Cresce, A. V.; Wang, F.; Yang, C.; Hu, Y.-S.; Xu, K.; Wang, C. “Water-in-

Salt” Electrolyte Makes Aqueous Sodium-Ion Battery Safe, Green, and Long-

Lasting. Advanced Energy Materials 2017, 7 (21), 1701189.

https://doi.org/10.1002/aenm.201701189.

174 References

(165) Zheng, Q.; Miura, S.; Miyazaki, K.; Ko, S.; Watanabe, E.; Okoshi, M.; Chou, C.-

P.; Nishimura, Y.; Nakai, H.; Kamiya, T.; Honda, T.; Akikusa, J.; Yamada, Y.;

Yamada, A. Sodium- and Potassium-Hydrate Melts Containing Asymmetric Imide

Anions for High-Voltage Aqueous Batteries. Angew. Chem. Int. Ed. 2019, 58 (40),

14202–14207. https://doi.org/10.1002/anie.201908830.

(166) Chen, H.; Zhang, Z.; Wei, Z.; Chen, G.; Yang, X.; Wang, C.; Du, F. Use of a Water-

in-Salt Electrolyte to Avoid Organic Material Dissolution and Enhance the

Kinetics of Aqueous Potassium Ion Batteries. Sustainable Energy Fuels 2020, 4

(1), 128–131. https://doi.org/10.1039/C9SE00545E.

(167) Jiang, L.; Lu, Y.; Zhao, C.; Liu, L.; Zhang, J.; Zhang, Q.; Shen, X.; Zhao, J.; Yu,

X.; Li, H.; Huang, X.; Chen, L.; Hu, Y.-S. Building Aqueous K-Ion Batteries for

Energy Storage. Nat Energy 2019. https://doi.org/10.1038/s41560-019-0388-0.

(168) Leonard, D. P.; Wei, Z.; Chen, G.; Du, F.; Ji, X. Water-in-Salt Electrolyte for

Potassium-Ion Batteries. ACS Energy Letters 2018, 3 (2), 373–374.

https://doi.org/10.1021/acsenergylett.8b00009.

(169) Liu, T.; Tang, L.; Luo, H.; Cheng, S.; Liu, M. A Promising Water-in-Salt

Electrolyte for Aqueous Based Electrochemical Energy Storage Cells with a Wide

Potential Window: Highly Concentrated HCOOK. Chem. Commun. 2019, 55 (85),

12817–12820. https://doi.org/10.1039/C9CC05927J.

(170) Cao, L.; Li, D.; Pollard, T.; Deng, T.; Zhang, B.; Yang, C.; Chen, L.; Vatamanu,

J.; Hu, E.; Hourwitz, M. J.; Ma, L.; Ding, M.; Li, Q.; Hou, S.; Gaskell, K.; Fourkas,

J. T.; Yang, X.-Q.; Xu, K.; Borodin, O.; Wang, C. Fluorinated Interphase Enables

Reversible Aqueous Zinc Battery Chemistries. Nat. Nanotechnol. 2021, 16 (8),

902–910. https://doi.org/10.1038/s41565-021-00905-4.

(171) Ni, Q.; Jiang, H.; Sandstrom, S.; Bai, Y.; Ren, H.; Wu, X.; Guo, Q.; Yu, D.; Wu,

C.; Ji, X. A Na3V2(PO4)2O1.6F1.4 Cathode of Zn‐Ion Battery Enabled by a Water‐

in‐Bisalt Electrolyte. Adv. Funct. Mater. 2020, 30 (36), 2003511.

https://doi.org/10.1002/adfm.202003511.

(172) Dong, Q.; Yao, X.; Zhao, Y.; Qi, M.; Zhang, X.; Sun, H.; He, Y.; Wang, D.

Cathodically Stable Li-O2 Battery Operations Using Water-in-Salt Electrolyte.

Chem 2018, 4 (6), 1345–1358. https://doi.org/10.1016/j.chempr.2018.02.015.

(173) Yang, C.; Suo, L.; Borodin, O.; Wang, F.; Sun, W.; Gao, T.; Fan, X.; Hou, S.; Ma,

Z.; Amine, K.; Xu, K.; Wang, C. Unique Aqueous Li-Ion/Sulfur Chemistry with

High Energy Density and Reversibility. Proc Natl Acad Sci USA 2017, 114 (24),

6197–6202. https://doi.org/10.1073/pnas.1703937114.

(174) Park, J. M.; Jana, M.; Nakhanivej, P.; Kim, B.-K.; Park, H. S. Facile Multivalent

Redox Chemistries in Water-in-Bisalt Hydrogel Electrolytes for Hybrid Energy

Storage Full Cells. ACS Energy Lett. 2020, 1054–1061.

https://doi.org/10.1021/acsenergylett.0c00059.

References 175

(175) Zhao, J.; Li, Y.; Peng, X.; Dong, S.; Ma, J.; Cui, G.; Chen, L. High-Voltage

Zn/LiMn0.8Fe0.2PO4 Aqueous Rechargeable Battery by Virtue of “Water-in-

Salt” Electrolyte. Electrochemistry Communications 2016, 69, 6–10.

https://doi.org/10.1016/j.elecom.2016.05.014.

(176) Wang, F.; Borodin, O.; Gao, T.; Fan, X.; Sun, W.; Han, F.; Faraone, A.; Dura, J.

A.; Xu, K.; Wang, C. Highly Reversible Zinc Metal Anode for Aqueous Batteries.

Nature Materials 2018, 17 (6), 543–549. https://doi.org/10.1038/s41563-018-

0063-z.

(177) Yang, C.; Chen, J.; Ji, X.; Pollard, T. P.; Lü, X.; Sun, C.-J.; Hou, S.; Liu, Q.; Liu,

C.; Qing, T.; Wang, Y.; Borodin, O.; Ren, Y.; Xu, K.; Wang, C. Aqueous Li-Ion

Battery Enabled by Halogen Conversion–Intercalation Chemistry in Graphite.

Nature 2019, 569 (7755), 245–250. https://doi.org/10.1038/s41586-019-1175-6.

(178) Guo, Q.; Kim, K.-I.; Li, S.; Scida, A. M.; Yu, P.; Sandstrom, S. K.; Zhang, L.; Sun,

S.; Jiang, H.; Ni, Q.; Yu, D.; Lerner, M. M.; Xia, H.; Ji, X. Reversible Insertion of

I–Cl Interhalogen in a Graphite Cathode for Aqueous Dual-Ion Batteries. ACS

Energy Lett. 2021, 6 (2), 459–467. https://doi.org/10.1021/acsenergylett.0c02575.

(179) Wu, X.; Xu, Y.; Zhang, C.; Leonard, D. P.; Markir, A.; Lu, J.; Ji, X. Reverse Dual-

Ion Battery via a ZnCl2 Water-in-Salt Electrolyte. J. Am. Chem. Soc. 2019, 141

(15), 6338–6344. https://doi.org/10.1021/jacs.9b00617.

(180) Forero-Saboya, J.; Hosseini-Bab-Anari, E.; Abdelhamid, M. E.; Moth-Poulsen, K.;

Johansson, P. Water-in-Bisalt Electrolyte with Record Salt Concentration and

Widened Electrochemical Stability Window. J. Phys. Chem. Lett. 2019, 10 (17),

4942–4946. https://doi.org/10.1021/acs.jpclett.9b01467.

(181) Ko, S.; Yamada, Y.; Miyazaki, K.; Shimada, T.; Watanabe, E.; Tateyama, Y.;

Kamiya, T.; Honda, T.; Akikusa, J.; Yamada, A. Lithium-Salt Monohydrate Melt:

A Stable Electrolyte for Aqueous Lithium-Ion Batteries. Electrochemistry

Communications 2019, 104, 106488.

https://doi.org/10.1016/j.elecom.2019.106488.

(182) Reber, D.; Grissa, R.; Becker, M.; Kühnel, R.; Battaglia, C. Anion Selection

Criteria for Water‐in‐Salt Electrolytes. Adv. Energy Mater. 2021, 11 (5), 2002913.

https://doi.org/10.1002/aenm.202002913.

(183) Chen, L.; Zhang, J.; Li, Q.; Vatamanu, J.; Ji, X.; Pollard, T. P.; Cui, C.; Hou, S.;

Chen, J.; Yang, C.; Ma, L.; Ding, M. S.; Garaga, M.; Greenbaum, S.; Lee, H.-S.;

Borodin, O.; Xu, K.; Wang, C. A 63 m Superconcentrated Aqueous Electrolyte for

High-Energy Li-Ion Batteries. ACS Energy Lett. 2020, 5 (3), 968–974.

https://doi.org/10.1021/acsenergylett.0c00348.

(184) Lukatskaya, M. R.; Feldblyum, J. I.; Mackanic, D. G.; Lissel, F.; Michels, D. L.;

Cui, Y.; Bao, Z. Concentrated Mixed Cation Acetate “Water-in-Salt” Solutions as

Green and Low-Cost High Voltage Electrolytes for Aqueous Batteries. Energy

Environ. Sci. 2018, 11 (10), 2876–2883. https://doi.org/10.1039/C8EE00833G.

176 References

(185) Becker, M.; Reber, D.; Aribia, A.; Battaglia, C.; Kühnel, R.-S. Hybrid Ionic

Liquid/Water-in-Salt Electrolytes Enable Stable Cycling of LTO/NMC811 Full

Cells. 13.

(186) Reber, D.; Takenaka, N.; Kühnel, R.-S.; Yamada, A.; Battaglia, C. Impact of Anion

Asymmetry on Local Structure and Supercooling Behavior of Water-in-Salt

Electrolytes. J. Phys. Chem. Lett. 2020, 11 (12), 4720–4725.

https://doi.org/10.1021/acs.jpclett.0c00806.

(187) Becker, M.; Kühnel, R.-S.; Battaglia, C. Water-in-Salt Electrolytes for Aqueous

Lithium-Ion Batteries with Liquidus Temperatures Below −10 °C. Chem.

Commun. 2019, 55 (80), 12032–12035. https://doi.org/10.1039/C9CC04495G.

(188) Nian, Q.; Zhang, X.; Feng, Y.; Liu, S.; Sun, T.; Zheng, S.; Ren, X.; Tao, Z.; Zhang,

D.; Chen, J. Designing Electrolyte Structure to Suppress Hydrogen Evolution

Reaction in Aqueous Batteries. ACS Energy Lett. 2021, 6 (6), 2174–2180.

https://doi.org/10.1021/acsenergylett.1c00833.

(189) Cao, L.; Li, D.; Hu, E. Solvation Structure Design for Aqueous Zn Metal Batteries.

Journal of the American Chemical Society 6.

(190) Hou, Z.; Dong, M.; Xiong, Y.; Zhang, X.; Zhu, Y.; Qian, Y. Formation of Solid–

Electrolyte Interfaces in Aqueous Electrolytes by Altering Cation‐Solvation Shell

Structure. Adv. Energy Mater. 2020, 1903665.

https://doi.org/10.1002/aenm.201903665.

(191) Wang, F.; Borodin, O.; Ding, M. S.; Gobet, M.; Vatamanu, J.; Fan, X.; Gao, T.;

Eidson, N.; Liang, Y.; Sun, W.; Greenbaum, S.; Xu, K.; Wang, C. Hybrid

Aqueous/Non-Aqueous Electrolyte for Safe and High-Energy Li-Ion Batteries.

Joule 2018, 2 (5), 927–937. https://doi.org/10.1016/j.joule.2018.02.011.

(192) Chen, J.; Vatamanu, J.; Xing, L.; Borodin, O.; Chen, H.; Guan, X.; Liu, X.; Xu,

K.; Li, W. Improving Electrochemical Stability and Low‐Temperature

Performance with Water/Acetonitrile Hybrid Electrolytes. Adv. Energy Mater.

2020, 10 (3), 1902654. https://doi.org/10.1002/aenm.201902654.

(193) Shang, Y.; Chen, N.; Li, Y.; Chen, S.; Lai, J.; Huang, Y.; Qu, W.; Wu, F.; Chen,

R. An “Ether‐In‐Water” Electrolyte Boosts Stable Interfacial Chemistry for

Aqueous Lithium‐Ion Batteries. Adv. Mater. 2020, 32 (40), 2004017.

https://doi.org/10.1002/adma.202004017.

(194) Hou, X.; Wang, R.; He, X.; Pollard, T. P.; Ju, X.; Du, L.; Paillard, E.; Frielinghaus,

H.; Barnsley, L. C.; Borodin, O.; Xu, K.; Winter, M.; Li, J. Stabilizing the Solid‐

Electrolyte Interphase with Polyacrylamide for High‐Voltage Aqueous Lithium‐

Ion Batteries. Angew. Chem. Int. Ed. 2021, anie.202107252.

https://doi.org/10.1002/anie.202107252.

(195) Jaumaux, P.; Yang, X.; Zhang, B.; Safaei, J.; Tang, X.; Zhou, D.; Wang, C.; Wang,

G. Localized Water‐In‐Salt Electrolyte for Aqueous Lithium‐Ion Batteries. Angew.

References 177

Chem. Int. Ed. 2021, 60 (36), 19965–19973.

https://doi.org/10.1002/anie.202107389.

(196) Liu, J.; Yang, C.; Chi, X.; Wen, B.; Wang, W.; Liu, Y. Water/Sulfolane Hybrid

Electrolyte Achieves Ultralow‐Temperature Operation for High‐Voltage Aqueous

Lithium‐Ion Batteries. Adv. Funct. Mater. 2021, 2106811.

https://doi.org/10.1002/adfm.202106811.

(197) Langevin, S. A.; Tan, B.; Freeman, A. W.; Gagnon, J. C.; Hoffman, C. M.; Logan,

M. W.; Maranchi, J. P.; Gerasopoulos, K. UV-Cured Gel Polymer Electrolytes with

Improved Stability for Advanced Aqueous Li-Ion Batteries. Chem. Commun. 2019,

55 (87), 13085–13088. https://doi.org/10.1039/C9CC06207F.

(198) Zhang, J.; Cui, C.; Wang, P.-F.; Li, Q.; Chen, L.; Han, F.; Jin, T.; Liu, S.;

Choudhary, H.; Raghavan, S. R.; Eidson, N.; von Cresce, A.; Ma, L.; Uddin, J.;

Addison, D.; Yang, C.; Wang, C. “Water-in-Salt” Polymer Electrolyte for Li-Ion

Batteries. Energy Environ. Sci. 2020, 13 (9), 2878–2887.

https://doi.org/10.1039/D0EE01510E.

(199) Cresce, A. Gel Electrolyte for a 4V Flexible Aqueous Lithium-Ion Battery. Journal

of Power Sources 2020, 10.

(200) He, X.; Yan, B.; Zhang, X.; Liu, Z.; Bresser, D.; Wang, J.; Wang, R.; Cao, X.; Su,

Y.; Jia, H.; Grey, C. P.; Frielinghaus, H.; Truhlar, D. G.; Winter, M.; Li, J.; Paillard,

E. Fluorine-Free Water-in-Ionomer Electrolytes for Sustainable Lithium-Ion

Batteries. Nature Communications 2018, 9 (1). https://doi.org/10.1038/s41467-

018-07331-6.

(201) Xie, J.; Liang, Z.; Lu, Y.-C. Molecular Crowding Electrolytes for High-Voltage

Aqueous Batteries. Nat. Mater. 2020, 19 (9), 1006–1011.

https://doi.org/10.1038/s41563-020-0667-y.

(202) Huang, M.; Yang, J.; Zhen, S.; Wan, C.; Jiang, X.; Ju, X. Fabrication of High

Li:Water Molar Ratio Electrolytes for Lithium-Ion Batteries. Chinese Chemical

Letters 2021, 32 (2), 834–837. https://doi.org/10.1016/j.cclet.2020.05.008.

(203) Widstrom, M. D.; Borodin, O.; Ludwig, K. B.; Matthews, J. E.; Bhattacharyya, S.;

Garaga, M.; V. Cresce, A.; Jarry, A.; Erdi, M.; Wang, C.; Greenbaum, S.; Kofinas,

P. Water Domain Enabled Transport in Polymer Electrolytes for Lithium-Ion

Batteries. Macromolecules 2021, 54 (6), 2882–2891.

https://doi.org/10.1021/acs.macromol.0c01960.

(204) Yang, C.; Chen, J.; Qing, T.; Fan, X.; Sun, W.; von Cresce, A.; Ding, M. S.;

Borodin, O.; Vatamanu, J.; Schroeder, M. A.; Eidson, N.; Wang, C.; Xu, K. 4.0 V

Aqueous Li-Ion Batteries. Joule 2017, 1 (1), 122–132.

https://doi.org/10.1016/j.joule.2017.08.009.

178 References

(205) Wang, F.; Lin, C.-F.; Ji, X.; Rubloff, G. W.; Wang, C. Suppression of Hydrogen

Evolution at Catalytic Surfaces in Aqueous Lithium Ion Batteries. J. Mater. Chem.

A 2020, 8 (30), 14921–14926. https://doi.org/10.1039/D0TA05568A.

(206) Sakai, S.; Yamada, I.; Miyahara, Y.; Kondo, Y.; Yokoyama, Y.; Abe, T.; Miyazaki,

K. Surface-Modified Li4Ti5O12 in Highly Concentrated Aqueous Solutions for Use

in Aqueous Rechargeable Lithium Batteries. J. Electrochem. Soc. 2020, 167 (12),

120512. https://doi.org/10.1149/1945-7111/ababd3.

(207) Chen, L.; Cao, L.; Ji, X.; Hou, S.; Li, Q.; Chen, J.; Yang, C.; Eidson, N.; Wang, C.

Enabling Safe Aqueous Lithium Ion Open Batteries by Suppressing Oxygen

Reduction Reaction. Nat Commun 2020, 1 (1), 2638.

https://doi.org/10.1038/s41467-020-16460-w.

(208) Hou, X.; Ju, X.; Zhao, W.; Wang, J.; He, X.; Du, L.; Yan, B.; Li, J.; Paillard, E.;

Sun, J.; Lin, H.; Winter, M.; Li, J. TiO2@LiTi2(PO4)3 Enabling Fast and Stable

Lithium Storage for High Voltage Aqueous Lithium-Ion Batteries. Journal of

Power Sources 2021, 484, 229255.

https://doi.org/10.1016/j.jpowsour.2020.229255.

(209) Subramanya, U. Carbon-Based Artificial SEI Layers for Aqueous Lithium-Ion

Battery Anodes. RSC Advances 2020, 8.

(210) Coustan, L.; Zaghib, K.; Bélanger, D. New Insight in the Electrochemical

Behaviour of Stainless Steel Electrode in Water-in-Salt Electrolyte. Journal of

Power Sources 2018, 399, 299–303.

https://doi.org/10.1016/j.jpowsour.2018.07.114.

(211) Hong, W. T.; Risch, M.; Stoerzinger, K. A.; Grimaud, A.; Suntivich, J.; Shao-Horn,

Y. Toward the Rational Design of Non-Precious Transition Metal Oxides for

Oxygen Electrocatalysis. Energy Environ. Sci. 2015, 8 (5), 1404–1427.

https://doi.org/10.1039/C4EE03869J.

(212) Dubouis, N.; Serva, A.; Salager, E.; Deschamps, M.; Salanne, M.; Grimaud, A.

The Fate of Water at the Electrochemical Interfaces: Electrochemical Behavior of

Free Water vs. Coordinating Water. 7.

(213) Suo, L.; Han, F.; Fan, X.; Liu, H.; Xu, K.; Wang, C. “Water-in-Salt” Electrolytes

Enable Green and Safe Li-Ion Batteries for Large Scale Electric Energy Storage

Applications. J. Mater. Chem. A 2016, 4 (17), 6639–6644.

https://doi.org/10.1039/C6TA00451B.

(214) Kunduraci, M.; Mutlu, R. N.; Gizir, A. M. Electrochemical Behavior of

LiNi0.6Mn0.2Co0.2O2 Cathode in Different Aqueous Electrolytes. Ionics 2020.

https://doi.org/10.1007/s11581-020-03490-z.

(215) Yang, C.; Rousse, G.; Louise Svane, K.; Pearce, P. E.; Abakumov, A. M.;

Deschamps, M.; Cibin, G.; Chadwick, A. V.; Dalla Corte, D. A.; Anton Hansen,

H.; Vegge, T.; Tarascon, J.-M.; Grimaud, A. Cation Insertion to Break the

References 179

Activity/Stability Relationship for Highly Active Oxygen Evolution Reaction

Catalyst. Nat Commun 2020, 11 (1), 1378. https://doi.org/10.1038/s41467-020-

15231-x.

(216) Laszczynski, N.; Solchenbach, S.; Gasteiger, H. A.; Lucht, B. L. Understanding

Electrolyte Decomposition of Graphite/NCM811 Cells at Elevated Operating

Voltage. J. Electrochem. Soc. 2019, 166 (10), A1853–A1859.

https://doi.org/10.1149/2.0571910jes.

(217) Xiong, D. J.; Ellis, L. D.; Li, J.; Li, H.; Hynes, T.; Allen, J. P.; Xia, J.; Hall, D. S.;

Hill, I. G.; Dahn, J. R. Measuring Oxygen Release from Delithiated LiNixMnyCo1-

x-yO2 and Its Effects on the Performance of High Voltage Li-Ion Cells. J.

Electrochem. Soc. 2017, 164 (13), A3025–A3037.

https://doi.org/10.1149/2.0291713jes.

(218) Jung, R.; Metzger, M.; Maglia, F.; Stinner, C.; Gasteiger, H. A. Oxygen Release

and Its Effect on the Cycling Stability of LiNixMnyCozO2 (NMC) Cathode

Materials for Li-Ion Batteries. J. Electrochem. Soc. 2017, 164 (7), A1361–A1377.

https://doi.org/10.1149/2.0021707jes.

(219) Zheng, J.; Xiao, J.; Xu, W.; Chen, X.; Gu, M.; Li, X.; Zhang, J.-G. Surface and

Structural Stabilities of Carbon Additives in High Voltage Lithium Ion Batteries.

Journal of Power Sources 2013, 227, 211–217.

https://doi.org/10.1016/j.jpowsour.2012.11.038.

(220) Gambou-Bosca, A.; Bélanger, D. Effect of the Formulation of the Electrode on the

Pore Texture and Electrochemical Performance of the Manganese Dioxide-Based

Electrode for Application in a Hybrid Electrochemical Capacitor. J. Mater. Chem.

A 2014, 2 (18), 6463. https://doi.org/10.1039/c3ta14910b.

(221) Betz, J.; Bieker, G.; Meister, P.; Placke, T.; Winter, M.; Schmuch, R. Theoretical

versus Practical Energy: A Plea for More Transparency in the Energy Calculation

of Different Rechargeable Battery Systems. Adv. Energy Mater. 2019, 9 (6),

1803170. https://doi.org/10.1002/aenm.201803170.

(222) Naik, K. M.; Sampath, S. Cubic Mo6S8-Efficient Electrocatalyst Towards

Hydrogen Evolution Over Wide PH Range. Electrochimica Acta 2017, 252, 408–

415. https://doi.org/10.1016/j.electacta.2017.09.015.

(223) Prigge, C.; Mueller-Warmuth, W.; Gocke, E.; Schoellhorn, R. Metallike Behavior

of Lithium Intercalated in Molybdenum Cluster Chalcogenides. Chem. Mater.

1993, 5 (10), 1493–1498. https://doi.org/10.1021/cm00034a020.

(224) Gocke, E.; Schoellhorn, R.; Aselmann, G.; Mueller-Warmuth, W. Molybdenum

Cluster Chalcogenides Mo6X8: Intercalation of Lithium via Electron/Ion Transfer.

Inorg. Chem. 1987, 26 (11), 1805–1812. https://doi.org/10.1021/ic00258a034.

180 References

(225) Xu, K.; von Wald Cresce, A. Li+-Solvation/Desolvation Dictates Interphasial

Processes on Graphitic Anode in Li Ion Cells. J. Mater. Res. 2012, 27 (18), 2327–

2341. https://doi.org/10.1557/jmr.2012.104.

(226) Xu, K.; von Cresce, A.; Lee, U. Differentiating Contributions to “Ion Transfer”

Barrier from Interphasial Resistance and Li + Desolvation at Electrolyte/Graphite

Interface. Langmuir 2010, 26 (13), 11538–11543.

https://doi.org/10.1021/la1009994.

(227) Lee, J.-W.; Pyun, S.-I. Investigation of Lithium Transport through LiMn2O4 Film

Electrode in Aqueous LiNO3 Solution. Electrochimica Acta 2004, 49 (5), 753–761.

https://doi.org/10.1016/j.electacta.2003.09.029.

(228) Abe, T.; Sagane, F.; Ohtsuka, M.; Iriyama, Y.; Ogumi, Z. Lithium-Ion Transfer at

the Interface Between Lithium-Ion Conductive Ceramic Electrolyte and Liquid

Electrolyte-A Key to Enhancing the Rate Capability of Lithium-Ion Batteries. J.

Electrochem. Soc. 2005, 152 (11), A2151. https://doi.org/10.1149/1.2042907.

(229) Hu, J.; Guo, H.; Li, Y.; Wang, H.; Wang, Z.; Huang, W.; Yang, L.; Chen, H.; Lin,

Y.; Pan, F. Understanding Li-Ion Thermodynamic and Kinetic Behaviors in

Concentrated Electrolyte for the Development of Aqueous Lithium-Ion Batteries.

Nano Energy 2021, 89, 106413. https://doi.org/10.1016/j.nanoen.2021.106413.

(230) Ayeb, A.; Otten, W. M.; Mank, A. J. G.; Notten, P. H. L. The Hydrogen Evolution

and Oxidation Kinetics during Overdischarging of Sealed Nickel-Metal Hydride

Batteries. J. Electrochem. Soc. 2006, 153 (11), A2055.

https://doi.org/10.1149/1.2336993.

(231) Ortiz-Rodríguez, J. C.; Singstock, N. R.; Perryman, J. T.; Hyler, F. P.; Jones, S. J.;

Holder, A. M.; Musgrave, C. B.; Velázquez, J. M. Stabilizing Hydrogen

Adsorption through Theory-Guided Chalcogen Substitution in Chevrel-Phase

Mo6X8 (X=S, Se, Te) Electrocatalysts. ACS Appl. Mater. Interfaces 2020, 12 (32),

35995–36003. https://doi.org/10.1021/acsami.0c07207.

(232) Alonso-Vante, N.; Schubert, B.; Tributsch, H. Transition Metal Cluster Materials

for Multi-Electron Transfer Catalysis. Materials Chemistry and Physics 1989, 22

(3–4), 281–307. https://doi.org/10.1016/0254-0584(89)90002-3.

(233) Dubouis, N.; Yang, C.; Beer, R.; Ries, L.; Voiry, D.; Grimaud, A. Interfacial

Interactions as an Electrochemical Tool To Understand Mo-Based Catalysts for the

Hydrogen Evolution Reaction. ACS Catal. 2018, 8 (2), 828–836.

https://doi.org/10.1021/acscatal.7b03684.

(234) Utsunomiya, T.; Hatozaki, O.; Yoshimoto, N.; Egashira, M.; Morita, M. Influence

of Particle Size on the Self-Discharge Behavior of Graphite Electrodes in Lithium-

Ion Batteries. Journal of Power Sources 2011, 196 (20), 8675–8682.

https://doi.org/10.1016/j.jpowsour.2011.06.070.

References 181

(235) Yazami, R.; Ozawa, Y. A Kinetics Study of Self-Discharge of Spinel Electrodes in

Li/LixMn2O4 Cells. Journal of Power Sources 2006, 153 (2), 251–257.

https://doi.org/10.1016/j.jpowsour.2005.10.012.

(236) Ozawa, Y.; Yazami, R.; Fultz, B. Self-Discharge Study of LiCoO2 Cathode

Materials. Journal of Power Sources 2003, 119–121, 918–923.

https://doi.org/10.1016/S0378-7753(03)00227-1.

(237) Notten, P. H. L.; Verbitskiy, E.; Kruijt, W. S.; Bergveld, H. J. Oxygen Evolution

and Recombination Kinetics Inside Sealed Rechargeable, Ni-Based Batteries. J.

Electrochem. Soc. 2005, 152 (7), A1423. https://doi.org/10.1149/1.1921849.

(238) Bullock, K. R.; McClelland, D. H. The Kinetics of the Self‐Discharge Reaction in

a Sealed Lead‐Acid Cell. J. Electrochem. Soc. 1976, 123 (3), 327–331.

https://doi.org/10.1149/1.2132819.

(239) Leblanc, P.; Blanchard, P.; Senyarich, S. Self‐Discharge of Sealed Nickel—Metal

Hydride Batteries: Mechanisms and Improvements. J. Electrochem. Soc. 1998, 145

(3), 844–847. https://doi.org/10.1149/1.1838355.

(240) Peled, E.; Golodnitsky, D.; Ardel, G. Advanced Model for Solid Electrolyte

Interphase Electrodes in Liquid and Polymer Electrolytes. J. Electrochem. Soc.

1997, 144 (8), L208–L210. https://doi.org/10.1149/1.1837858.

(241) Gauthier, M.; Carney, T. J.; Grimaud, A.; Giordano, L.; Pour, N.; Chang, H.-H.;

Fenning, D. P.; Lux, S. F.; Paschos, O.; Bauer, C.; Maglia, F.; Lupart, S.; Lamp,

P.; Shao-Horn, Y. Electrode–Electrolyte Interface in Li-Ion Batteries: Current

Understanding and New Insights. J. Phys. Chem. Lett. 2015, 6 (22), 4653–4672.

https://doi.org/10.1021/acs.jpclett.5b01727.

(242) Jones, J.; Anouti, M.; Caillon-Caravanier, M.; Willmann, P.; Lemordant, D.

Thermodynamic of LiF Dissolution in Alkylcarbonates and Some of Their

Mixtures with Water. Fluid Phase Equilibria 2009, 285 (1–2), 62–68.

https://doi.org/10.1016/j.fluid.2009.07.020.

(243) Tasaki, K.; Goldberg, A.; Lian, J.-J.; Walker, M.; Timmons, A.; Harris, S. J.

Solubility of Lithium Salts Formed on the Lithium-Ion Battery Negative Electrode

Surface in Organic Solvents. J. Electrochem. Soc. 2009, 156 (12), A1019.

https://doi.org/10.1149/1.3239850.

(244) Strmcnik, D.; Castelli, I. E.; Connell, J. G.; Haering, D.; Zorko, M.; Martins, P.;

Lopes, P. P.; Genorio, B.; Østergaard, T.; Gasteiger, H. A.; Maglia, F.;

Antonopoulos, B. K.; Stamenkovic, V. R.; Rossmeisl, J.; Markovic, N. M.

Electrocatalytic Transformation of HF Impurity to H2 and LiF in Lithium-Ion

Batteries. Nat Catal 2018, 1 (4), 255–262. https://doi.org/10.1038/s41929-018-

0047-z.

(245) Vincze, J.; Valiskó, M.; Boda, D. The Nonmonotonic Concentration Dependence

of the Mean Activity Coefficient of Electrolytes Is a Result of a Balance between

182 References

Solvation and Ion-Ion Correlations. The Journal of Chemical Physics 2010, 133

(15), 154507. https://doi.org/10.1063/1.3489418.

(246) McEldrew, M.; Goodwin, Z. A. H.; Bi, S.; Bazant, M. Z.; Kornyshev, A. A. Theory

of Ion Aggregation and Gelation in Super-Concentrated Electrolytes. J. Chem.

Phys. 2020, 152 (23), 234506. https://doi.org/10.1063/5.0006197.

(247) Liu, D.; Yu, Q.; Liu, S.; Qian, K.; Wang, S.; Sun, W.; Yang, X.-Q.; Kang, F.; Li,

B. Evolution of Solid Electrolyte Interface on TiO2 Electrodes in an Aqueous Li-

Ion Battery Studied Using Scanning Electrochemical Microscopy. J. Phys. Chem.

C 2019, acs.jpcc.9b01412. https://doi.org/10.1021/acs.jpcc.9b01412.

(248) Zhang, H.; Wang, D.; Shen, C. In-Situ EC-AFM and Ex-Situ XPS Characterization

to Investigate the Mechanism of SEI Formation in Highly Concentrated Aqueous

Electrolyte for Li-Ion Batteries. Applied Surface Science 2020, 507, 145059.

https://doi.org/10.1016/j.apsusc.2019.145059.

(249) He, M.; Guo, R.; Hobold, G. M.; Gao, H.; Gallant, B. M. The Intrinsic Behavior of

Lithium Fluoride in Solid Electrolyte Interphases on Lithium. Proc Natl Acad Sci

USA 2020, 117 (1), 73–79. https://doi.org/10.1073/pnas.1911017116.

(250) Aurbach, D.; Ein‐Ely, Y.; Zaban, A. The Surface Chemistry of Lithium Electrodes

in Alkyl Carbonate Solutions. J. Electrochem. Soc. 1994, 141 (1), L1–L3.

https://doi.org/10.1149/1.2054718.

(251) Guo, R.; Wang, D.; Zuin, L.; Gallant, B. M. Reactivity and Evolution of Ionic

Phases in the Lithium Solid–Electrolyte Interphase. ACS Energy Lett. 2021, 6 (3),

877–885. https://doi.org/10.1021/acsenergylett.1c00117.

(252) Zhang, X.-Q.; Cheng, X.-B.; Chen, X.; Yan, C.; Zhang, Q. Fluoroethylene

Carbonate Additives to Render Uniform Li Deposits in Lithium Metal Batteries.

Adv. Funct. Mater. 2017, 27 (10), 1605989.

https://doi.org/10.1002/adfm.201605989.

(253) Aurbach, D. A Short Review of Failure Mechanisms of Lithium Metal and

Lithiated Graphite Anodes in Liquid Electrolyte Solutions. Solid State Ionics 2002,

148 (3–4), 405–416. https://doi.org/10.1016/S0167-2738(02)00080-2.

(254) Li, C.; Lan, Q.; Yang, Y.; Shao, H.; Zhan, H. Flexible Artificial Solid Electrolyte

Interphase Formed by 1,3-Dioxolane Oxidation and Polymerization for Metallic

Lithium Anodes. ACS Appl. Mater. Interfaces 2019, 11 (2), 2479–2489.

https://doi.org/10.1021/acsami.8b16080.

(255) Wrogemann, J. M.; Künne, S.; Heckmann, A.; Rodríguez‐Pérez, I. A.; Siozios, V.;

Yan, B.; Li, J.; Winter, M.; Beltrop, K.; Placke, T. Development of Safe and

Sustainable Dual‐Ion Batteries Through Hybrid Aqueous/Nonaqueous

Electrolytes. Adv. Energy Mater. 2020, 10 (8), 1902709.

https://doi.org/10.1002/aenm.201902709.

References 183

(256) Zhang, X.; Dong, M.; Xiong, Y.; Hou, Z.; Ao, H.; Liu, M.; Zhu, Y.; Qian, Y.

Aqueous Rechargeable Li+/Na+ Hybrid Ion Battery with High Energy Density and

Long Cycle Life. Small 2020, 16 (41), 2003585.

https://doi.org/10.1002/smll.202003585.

(257) Miyazaki, K.; Takenaka, N.; Watanabe, E.; Iizuka, S.; Yamada, Y.; Tateyama, Y.;

Yamada, A. First-Principles Study on the Peculiar Water Environment in a

Hydrate-Melt Electrolyte. J. Phys. Chem. Lett. 2019, 10 (20), 6301–6305.

https://doi.org/10.1021/acs.jpclett.9b02207.

(258) Nikitina, V. A.; Zakharkin, M. V.; Vassiliev, S. Yu.; Yashina, L. V.; Antipov, E.

V.; Stevenson, K. J. Lithium Ion Coupled Electron-Transfer Rates in

Superconcentrated Electrolytes: Exploring the Bottlenecks for Fast Charge-

Transfer Rates with LiMn2O4 Cathode Materials. Langmuir 2017, 33 (37), 9378–

9389. https://doi.org/10.1021/acs.langmuir.7b01016.

(259) Brouillette, D.; Perron, G.; Desnoyers, J. E. Apparent Molar Volume, Heat

Capacity, and Conductance of Lithium Bis(Trifluoromethylsulfone)Imide in

Glymes and Other Aprotic Solvents. Canadian Journal of Physics 1997, No. 75,

1608–1614.

(260) Marković, N. M.; Grgur, B. N.; Ross, P. N. Temperature-Dependent Hydrogen

Electrochemistry on Platinum Low-Index Single-Crystal Surfaces in Acid

Solutions. J. Phys. Chem. B 1997, 101 (27), 5405–5413.

https://doi.org/10.1021/jp970930d.

(261) Tang, Y.; Zhang, Y.; Li, W.; Ma, B.; Chen, X. Rational Material Design for

Ultrafast Rechargeable Lithium-Ion Batteries. Chem. Soc. Rev. 2015, 44 (17),

5926–5940. https://doi.org/10.1039/C4CS00442F.

(262) Yoo, H. D.; Markevich, E.; Salitra, G.; Sharon, D.; Aurbach, D. On the Challenge

of Developing Advanced Technologies for Electrochemical Energy Storage and

Conversion. Materials Today 2014, 17 (3), 110–121.

https://doi.org/10.1016/j.mattod.2014.02.014.

(263) Köhler, U.; Antonius, C.; Bäuerlein, P. Advances in Alkaline Batteries. Journal of

Power Sources 2004, 127 (1–2), 45–52.

https://doi.org/10.1016/j.jpowsour.2003.09.006.

(264) Tarascon, J.-M.; Gozdz, A. S.; Schmutz, C.; Shokoohi, F.; Warren, P. C.

Performance of Bellcore’s Plastic Rechargeable Li-Ion Batteries. Solid State Ionics

1996, 86–88, 49–54. https://doi.org/10.1016/0167-2738(96)00330-X.

(265) Lepoivre, F.; Grimaud, A.; Larcher, D.; Tarascon, J.-M. Long-Time and Reliable

Gas Monitoring in Li-O 2 Batteries via a Swagelok Derived Electrochemical Cell.

J. Electrochem. Soc. 2016, 163 (6), A923–A929.

https://doi.org/10.1149/2.0421606jes.

184 References

(266) Lepoivre, F. Study and Improvement of Non-Aqueous Lithium-Air Batteries via

the Development of a Silicon-Based Anode. Thesis manuscript 2016.

(267) Wei Yin. Fundamental Understanding of High-Energy-Density Li-Ion Batteries

and beyond : Structure-Property Relationships and Reaction Chemistries. Thesis

manuscript 2019.

(268) Yin, W.; Mariyappan, S.; Grimaud, A.; Tarascon, J. M. Rotating Ring Disk

Electrode for Monitoring the Oxygen Release at High Potentials in Li-Rich

Layered Oxides. J. Electrochem. Soc. 2018, 165 (14), A3326–A3333.

https://doi.org/10.1149/2.0481814jes.

(269) Jones, J.; Anouti, M.; Caillon-Caravanier, M.; Willmann, P.; Sizaret, P.-Y.;

Lemordant, D. Solubilization of SEI Lithium Salts in Alkylcarbonate Solvents.

Fluid Phase Equilibria 2011, 305 (2), 121–126.

https://doi.org/10.1016/j.fluid.2011.03.007.

(270) Jones, J.; Anouti, M.; Caillon-Caravanier, M.; Willmann, P.; Lemordant, D.

Lithium Fluoride Dissolution Equilibria in Cyclic Alkylcarbonates and Water.

Journal of Molecular Liquids 2010, 153 (2–3), 146–152.

https://doi.org/10.1016/j.molliq.2010.02.006.

(271) Han, K. S.; Yu, Z.; Wang, H.; Redfern, P. C.; Ma, L.; Cheng, L.; Chen, Y.; Hu, J.

Z.; Curtiss, L. A.; Xu, K.; Murugesan, V.; Mueller, K. T. Origin of Unusual Acidity

and Li + Diffusivity in a Series of Water-in-Salt Electrolytes. J. Phys. Chem. B

2020, 124 (25), 5284–5291. https://doi.org/10.1021/acs.jpcb.0c02483.

APPENDIX

- APPENDIX - CHAPTER 1

Table A.1. 1 Summary of the main salts listed in Chapter 1.

Name Abbreviatio

n Chemical formula

Lithium hexafluorophosphate LiPF6

Lithium hexafluoroarsenate LiAsF6

Lithium acetate LiAc

Lithium triflouromethanesulfonate or

Lithium triflate LiOtf

Lithium bis(fluorosulfonyl)imide LiFSI

Lithium bis(trifluoromethanesulfonyl)imide LiTFSI

Lithium bis(pentafluoroethanesulfonyl)imide

LiBETI

Lithium pentafluoroethanesulfonyl) (trifluoromethanesulfonyl)imide

LiPTFSI

Lithium

(fluorosulfonyl)(trifluoromethylsulfonyl)imide

LiFTFSI

Lithium propylsulfonate-butylsulfonate-dimethyl-amine

LiMM3411

1-Ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide

EMim TFSI

Appendix 187

Table A.1. 2 Summary of the main solvents listed in Chapter 1.

Name Abbreviation Chemical formula

1,3-dioxolane DOL

1,5-pentanediol PD

2-fluoroethyl methyl carbonate FEMC

Acetonitrile ACN

Diethylcarbonate DEC

Dimethoxyethane DME

Dimethyl carbonate DMC

Dimethyl sulfoxide DMSO

Dimethylformamide DMF

Ethyl methyl carbonate EMC

Ethylene carbonate EC

Fluoroethylene carbonate FEC

Polyacrylamide PAM

Propylene carbonate PC

Sulfolane SL

Tetraethylene glycol dimethyl ether TEGDME

Triethyl phosphate TEP

Trimethyl phosphate TMP

Urea Urea

Water H2O

188 Appendix

SEI formation mechanisms in Water-in-salt electrolyte

The first SEI formation mechanism reported relies on the direct electrochemical

reduction of anion or anion clusters, such as Li2(TFSI)+146.. Alike in organic

superconcentrated electrolyte, anions would partially donate electrons to Li+ to form

Li2(TFSI)+ complex, therefore lowering the anion LUMO and increasing its reduction

potential43,45,54, as revealed by quantum chemistry calculation146 (see Figure A.1. 1a).

Similarly, the electronic structure of water isolated monomers or clusters changes,

leading to an inversion of the LUMO levels between water and anion, thus promoting

anion reduction at low potential257. In addition, Li2CO3 and Li2O were determined as

contributors due to the complementary reduction of CO2 and O2 dissolved gases in the

electrolyte146 (Equation A.1. 1 and Equation A.1. 2). However, as discussed in section

1.2, the determination of HOMO/LUMO energy level is not sufficient to explain changes

in interfacial reactivity. Therefore, other SEI mechanisms were discussed in literature.

Dubouis et al.134 observed that the HER occurs at potential above 2 V vs Li+/Li in WiSE,

and that TFSI- anions chemically degrade in the strong alkaline environment produced

by HER (see Figure A.1. 1b). Based on this mechanism, LiOH and CFx elements were

found to contribute to the SEI composition. Eventually, complementary to the chemical

degradation path, Bouchal et al.149 suggested that a dynamic precipitation/dissolution

of LiTFSI salt also contributes to the SEI, as also reported by Nikitina et al.258.

Moreover, the SEI formation may be influence by the preferential adsorption of Li+

on the electrode surface. For instance, the strong Li+ adsorption on Pt surface brings

more water to the interface154 promoting water reduction as shown by observing almost

no changes in HER onset potential on Pt compared to Al, enabling a downshift of HER. A

part from the HER catalysis related to the electrode material and the Li+ cation,

hypothesis based on MD simulations proposed that the electrode material also

influences the anion orientation, thus modifying the SEI formation154. However, these

hypothesis remain under debate.

Appendix 189

Figure A.1. 1 (a) Predicted reduction potentials from quantum chemistry calculations. Adapted from Ref111. (b) Schematic illustration of the formation of the SEI following a ‘‘water reduction mediated mechanism’’ occurring in 20 m LiTFSI WiSE. Adapted from Ref134.

Equation A.1. 1: Electrochemical reduction of dissolved O2 (1) and CO2 (2) contributing to SEI formation.

𝑂2 + 4 ∙ 𝐿𝑖+ + 4 ∙ 𝑒− = 2 ∙ 𝐿𝑖2𝑂 (1)

2 ∙ 𝐶𝑂2 + 𝑂2 + 2 ∙ 𝐿𝑖+ + 2 ∙ 𝑒− = 𝐿𝑖2𝐶𝑂3 (2)

Equation A.1. 2: Chemical reaction contributing to SEI formation. 𝐶𝑂2 + 𝐿𝑖2𝑂 = 𝐿𝑖2𝐶𝑂3

190 Appendix

Table A.1. 3 Summary of the polymer coating used to prevent HER.

PGE

Abbreviation

Name Chemical formula

Role and (mass ratio)

Polymer matrix

to electrol

yte (mass ratio)

UV-curable

HEA 2-hydroxyethyl acrylate

Monomer (89)

25

MPEGA poly(ethylene glycol) methyl ether acrylate

Co-monomer

(9)

PEGDA poly(ethylene glycol)

diacrylate

Cross-linker(2)

DMPA 2,2-dimethoxy-2-

phenylacetophenoe

Photo-initiator

(0.2 wt%)

Electrolyte

75 LiTFSI:H2O:TMP199 37:40:23 (molar ratio)

21 m LiTFSI : 7 m LiOtf197

Molecular

crowding

PEG/PEO poly(ethylene

glycol)/poly(ethylene oxide)

Electrolyte

2 m LiTFSI in PEG0.94(H20)0.06

- APPENDIX - CHAPTER 3

Energy density and specific energy as function of

electrolyte volume

Figure A.3. 1 Specific energy and energy density as function of the electrolyte volume

calculated thanks to the model developed by Betz et al.221.

192 Appendix

Cycling behavior in 3-electrodes set up

Figure A.3. 2 (a-c) Illustration of the protocol used to assess SEI stability over self-discharge cycling. Potential as function of time for 3 cells using a 3-electrodes cell based on Mo6S8 as counter electrode, LFP as working electrode and a ring of Li0.5FePO4 as reference in 20 m LiTFSI. (d-f) Nyquist plot obtained by electrochemical impedance spectroscopy of LFP determined over cycling. (i-k) Nyquist plot obtained by electrochemical impedance spectroscopy of Mo6S8 determined over cycling. Frequency range was comprised between 1 MHz and 50 mHz with a signal amplitude of 20 mV.

Three cells were assembled to assess SEI stability over cycling. Despite the poor

performances obtained when using a 3-electrodes cell, their cycling curves show a

rather good reproducibility. In addition, electrochemical impedance spectroscopy

assessed on LFP gives similar response (see Figure A.3. 2d-f). Such observations are

further confirmed by the MO6S8 impedance spectra (see Figure A.3. 2g-i) that are only

affected by the cell SOC and not the SEI evolution. Therefore, as shown in Figure A.3. 3

showing detailed EIS spectra as function of the cell SOC, the contribution of the LFP

counter electrode to the cell impedance does not vary throughout the experiment, thus

confirming that the high capacity loading of LFP:Mo6S8 (4:1) is enough to consider our

set-up as a pseudo-half-cell and to prevent LFP impedance to be impacted by the cell

Appendix 193

SOC. However, a clear inductive loop is shown on the LFP spectra. Issues in positioning

the electrode toward the reference electrode may explain such behavior.

Figure A.3. 3 Nyquist plot obtained by electrochemical impedance spectroscopy of LFP working electrode over cycling. Frequency range was comprised between 1 MHz and 50 mHz with a signal amplitude of 20 mV. (a) Summary of Nyquist plot obtained during cycling, (b) Nyquist plot obtained at charged state, (c) Nyquist plot of LFP at discharged state.

194 Appendix

Differential scanning calorimetry (DSC)

Figure A.3. 4 Differential scanning calorimetry (DSC) experiments performed at 2 °C/min between 35 °C and -60 °C upon cooling and back to 60 °C on heating. (a) Electrolytes with different concentrations used for calibration (19.803 m, 20.591 m, 20.910 m, 21.992, 22.9983 m), (b) zoom on the crystallization peak of calibration data.

To assess the impact of water consumption by recording the evolution of water

content in the electrolyte as function of cycle number and self-discharge period, a

calibration curve is determined. To do so, separators are soaked with electrolytes with

known concentrations and DSC experiments are performed (see Figure A.3. 4a and b).

A first exothermic peak is observed between 0 °C and 20 °C, shifting to greater

temperature with increasing concentration, which is attributed to the beginning of the

H2O∙LiTFSI crystallization126. To be consistent between melting of pure components and

the liquidus of transition of adjacent phases, phase transitions in WiSE are assessed at

peak temperature, rather than onset temperature, as reported by Ding et al.126. A

second peak at ≈ -20 °C following the crystallization of (H2O)4Li∙TFSI 126 is then observed.

These two peaks could be related to the liquidus (between 0 °C to 20 °C) and solidus (≈

-20 °C) temperature of 20 m LiTFSI electrolyte, as reported in the phase diagram

proposed by Ding et al.126 (see Figure 1. 17). Nonetheless, the associated temperatures

measured in our experiments differ from the one reported by Ding et al., with the solidus

temperature reported at -8.7 °C for LiTFSI concentration comprised between 13.9 m and

55.5 m, while we observed a solidus temperature at ≈ -20 °C. Finally, upon heating, an

endothermic “wave” that corresponds to the melting point of H2O∙LiTFSI is observed, in

agreement with previous DSC experiments carried out on 21 m LITFSI111,163. This wave

shifts to higher temperature with increase in concentration in line with the trend

observed for crystallization.

Appendix 195

Besides, for the 22.993 m electrolyte an exothermic peak is observed at -46.4 °C upon

cooling, followed by an endothermic peak upon heating at ≈ -5 °C (see Figure A.3. 4a).

Based on the shape of the -46.4 °C peak, one can hypothesize that it is related to the

crystallization of a pure component, presumably water. However, the corresponding

endothermic peak (≈ -5 °C) does not correspond to the melting of a pure component.

This behavior could be related to the incongruent melting of unstable solvates, as

reported by Perron et al.259. Similar peaks were also observed by Reber et al.163 for a 21

m LiTFSI solution but at temperature lower than -60 °C for the exothermic peak.

However, considering the 20.910 and 21.992 m electrolyte, we did not observe this

peak. The difference may be related to the use of mesocarbon microbeads to promote

crystallization previously employed by Reber et al.163.

Based on the results shown in Figure A.3. 4, the following fit is obtained:

𝑇𝑐𝑟𝑦𝑠𝑡𝑎𝑙𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 𝑝𝑒𝑎𝑘 = 4.68 ∙ 𝐶𝐿𝑖𝑇𝐹𝑆𝐼𝑚𝑜𝑙 𝑘𝑔⁄

− 91.5 𝑤𝑖𝑡ℎ 𝑅2 = 0.997.

196 Appendix

Activation energy of direct HER in 20 m LiTFSI

Pressure cell experiments

Figure A.3. 5 Potential (black line) and pressure evolution (red line) as function of time for a cell cycled at 0.15 C at 45 °C. Prior to the cycling step, the cell was pre-cycled at 1C and 25 °C for 10 cycles.

Figure A.3. 5 shows the pressure evolution as function of time for a cell which is

cycled at 45 °C with a C-rate of 0.15C. Prior to this cycling test, the cell underwent a pre-

cycling step of 10 cycles performed at 1C, 25 °C. Contrary to what is observed at 25 °C

or 35 °C, a pure HER plateau appears above 1.4 V. Indeed, due to the low C-rate and the

manually-applied pressure in Swagelok cell, parasitic reactions are favored when

increasing temperature thus explaining the appearance of such plateau at 45 °C and 55

°C.

Gas calculations made to determine the activation energy of HER on Mo6S8 take into

consideration the pure HER plateau to calculate H2 evolution. In addition, the hydrogen

rate (𝐿𝑛 (∆𝑃

𝑃0⁄

∆𝑡)) calculated at 55 °C using two cells give close values ranging from -7 to

-7.3 suggesting that the experiments are reproducible.

Appendix 197

Figure A.3. 6 Arrhenius plot obtained from cycling of Mo6S8/LFP pressure cell in 20 m LiTFSI as function of temperature. The logarithm of the hydrogen rate ((ΔP/P0)/Δtime) as function of the inverse of temperature. Hydrogen rate is calculated considering charge time (full line) or charge and discharge time (dash line).

Eventually, Figure A.3. 6 shows the Arrhenius plot considering either the charge time

or the charge and discharge time as the time during which hydrogen is produced. The

rather close value of the slope (-8.62 and -9.79) found tend to confirm that hydrogen is

majorly produced during charge thus further confirming the hypothesis used in this

study.

Electrochemical cell: HER on Pt

Temperature-controlled CV were performed in an electrochemical cell to determine

HER activation energy on Pt in 20 m LiTFSI; the corresponding polarization curves and

Tafel plots are shown in Figure A.3. 7a and b. Looking into the polarization curves,

limitations regarding the experiment can be observed since an overpotential of 600 mV

must be applied to reach the potential range in which HER occurs.

198 Appendix

Figure A.3. 7 (a) Cyclic voltammetry performed at 50 mV/s in 20 m LiTFSI in a 3-electrodes cell using Pt disk connected to a rotating disk electrode (ω=1600 rpm) as working electrode, Pt wire as counter electrode and a leakless electrode as reference electrode. (b) Tafel representation of the cyclic voltammetry.

The most accurate approach to extract the activation energy for the HER relies on

analysis made in the so-called micropolarization region260, at potentials very close to the

equilibrium potential in order to reduce effects related to mass transport. To assess the

kinetic parameters, one need to verify that all the mass-transport limitation is not

influencing the electrochemistry. The effect of mass transport can be quantified thanks

to Koutecky-Levich equation (see Equation A.3. 1).

Equation A.3. 1

1

𝑗𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 =

1

𝑗𝑘 + (

1

0.62𝑛𝐹𝐴𝐷−23 𝜈

16𝐶𝑏𝑢𝑙𝑘

)𝜔−1/2

with jk the kinetic current (A), n the number of electron transferred, F the Faraday

constant (C/mol), A the electrode area (m²), D the diffusion coefficient for the reactant

(m²/s), ν the kinematic viscosity (m²/s), Cbulk the concentration in the bulk (mol/m3) and

ω the rotation speed (rad/s).

Thus, in the absence of mass transport control, the second term of the equation is

constant regardless of the rotation speed of the RDE. Figure A.3. 8 shows the inverse of

the forward current density determined at an overpotential of -600 mV as function of

the inverse of the square root of the rotation speed. One can note that from 25 °C to

45 °C, the reaction is mix-controlled by mass transport and kinetics both limiting the

current in the 600 mV overpotential range studied in this work. At 55 °C, Koutecky-Levich

analysis shows less influence of mass transport on the current density. Besides, changes

Appendix 199

in the active surface area of the Pt electrode occur upon polarization, due to the SEI

formation and the surface passivation. This effect is observed by the crossover of the

curves between the forward and the backward scan (see Figure A.3. 7).

Figure A.3. 8 Koutecky-Levich analysis. Inverse of the forward current assess at an overpotential η = -600 mV as function of the inverse of the square root of the rotating speed of the rotating disk electrode (RDE). Four rotating speed were considered: 400 rpm, 600 rpm, 900 rpm and 1600 rpm. All measurements were conducted in a 3-electrodes electrochemical cell using Pt disk as working electrode, Pt wire as counter electrode and a leakless reference. Temperature was controlled to 25 °C, 35 °C, 45 °C and 55 °C.

Altogether, these parameters suggest that caution must be taken when discussing

the activation energy measured to be ≈ 10 kJ/mol at an overpotential (η) of -600 mV and

a rotation speed (ω) of 1600 rpm (see Figure A.3. 9). Indeed, 58 kJ/mol must be added

to obtain the activation energy at the equilibrium potential.

200 Appendix

Figure A.3. 9 Arrhenius analysis made at an overpotential of -600 mV, a rotating speed of 1600 rpm. All measurements were conducted in a 3-electrodes electrochemical cell using Pt disk connected to a rotating disk electrode as working electrode, Pt wire as counter electrode and a leakless reference in 20 m LiTFSI electrolyte. Temperature was controlled to 25 °C, 35 °C, 45 °C and 55 °C.

Spider chart data

Data used to make the spider chart shown in Figure 3. 14 are either directly taken

from literature or experimentally assess (in the case of WiSE-based LIB) and then

calculated thanks to the model develop by Betz et al. 221 . Table A.3. 1, Table A.3. 2 and

Table A.3. 3 summarize the data used to draw the comparative spider chart. Rate

capability of Lead-acid, Ni-Cd, Ni-MH and LIB devices were extracted and calculated from

Ref8,221,261,262.

Appendix 201

Table A.3. 1 Summary of specific energy and energy density as function of the technology.

Specific energy (Wh/kgcell)

Energy density (Wh/Lcell)

Technology Devices Min Max Min Min

Lead-acid Grid/

Vehicles 30 9 60 5,6 80 6 130 6

Ni-Cd Portable/Vehicles

30 7 60 8 55 7 150 263

Ni-MH Portable/Vehicles

42 5 110 5 135 7 275 263

NMC111/graphite adapted from

Ref221 18650 196 422

Superconcentrated aqueous

adapted from Ref221

18650 54

Mo6S8/LFP (WiSE)

99 LTO/NMC

(WiBS)

161 Mo6S8/LFP

(WiSE)

247 LTO/NMC

(WiBS)

Table A.3. 2 Summary of Self-discharge rate as function of the technology.

Technology Self-discharge rate (%/month)

Ref 7 Min Max

Lead-acid 1 15

Ni-Cd 5 20

Ni-MH 15 30

NMC111/graphite 2 10

20m LiTFSI (experimental)

30 75

202 Appendix

Table A.3. 3 Summary of the operating temperature range as function of the technology.

Operating temperature (°C)

Technology Min Max

Lead-acid -40 5 60 5

Ni-Cd -20 7 70 7

Ni-MH -30 5 65 5

NMC111/graphite -20 5 60 5

20m LiTFSI (experimental)

20 40

Table A.3. 4 Summary of the energy efficiency as function of the technology.

Technology Energy efficiency Min Min

Lead-acid 70 5 85 6

Ni-Cd 72,5 8 85 6

Ni-MH 70 8 90 7

NMC111/graphite 90 6 95 5

20m LiTFSI (experimental)

92 92

In their study, Betz et al.221 split the energy density and specific energy calculations

in six steps from a theoretical point (Step 1) of view to a practical one (from Step 2 to

Step 6) as described in Figure A.3. 10.

Table A.3. 5, Table A.3. 6, Table A.3. 7, Table A.3. 8, Table A.3. 9 and Table A.3. 10

detail the parameter and calculations made thanks to the model developed by

Appendix 203

Figure A.3. 10 Illustration of the steps used in Ref. 221 to determine the energy density and specific energy of a LIB cell as function of the influence of the different cell component.

204 Appendix

Table A.3. 5 Energy density of 5 LIB as function of the step considered in calculation.

Table A.3. 6 Specific energy of 5 LIB as function of the step considered in calculation.

Table A.3. 7 Active materials parameter taken into account the model.

Table A.3. 8 Cell parameters, inactive cell component and housing properties used in the LIB model by the 6 steps of calculations.

Appendix 205

Table A.3. 9 Electrolyte properties used to describe a LIB cell in the model.

Table A.3. 10 Summary of the battery system calculation made with the model.

MATERIALS & METHODS

208 Materials & Methods

Material preparation

LiFePO4 (LFP) and LiNi0.6Mn0.2Co0.202 (NMC622) were purchased from Umicore.

LiTi2(PO4)3 (LTP) was prepared by solid-state reaction of stoichiometric amounts of

Li2CO3 (>99 %, Sigma Aldrich), TiO2 (>99 %, Sigma Aldrich) and NH4H2PO4 (98 %, Alfa

Aesar). The precursors were grinded and heated at 200 °C for 2 h and finally 930 °C for

24 h in air. Mo6S8 was either obtained from ISCR (Institut des Sciences Chimiques de

Rennes) or synthetized using the following protocol. Solid-state reaction was carried out

by grinding and heating to 1050 °C (2 °C/min) for 72 h in a sealed tube stoichiometric

amounts of Cu, Mo (99.95 %, Alfa Aesar) and S (99.98 %, Sigma Aldrich). An excess of

3.7 % (molar ratio) in Cu was then added and tube vacuum sealed and heated at 1050 °C

for 24 h (2 °C/min). When MoS2 impurities were detected, the as-synthetized powder

was treated under H2 reductive atmosphere in a boat crucible for 12 h (5 °C/min). The

resulting sample was then acid-leached overnight in HCl 6 M under oxygen bubbling.

The powder and the supernatant were separated by centrifugation and the samples

were washed with distilled water until pH=7, prior to being dried at 80 °C under vacuum

overnight.

X-ray diffraction (XRD) was performed to confirm the LTP and Mo6S8 phases purity

using a BRUKER D8 Advance diffractometer with Cu Kα radiation (λKα1 = 1.54056 Å, λKα2

= 1.54439 Å).

1.2.1 Bellcore technique

For battery assembly, Mo6S8, NMC622, LFP and LTP composite self-standing electrodes

are fabricated using Bellcore technique264. Active materials (AM), Carbon super P (Csp,

Timcal) and PVdF-HFP (Poly(vinylidenefluoride-hexafluoropropylene, Solvay) are

grinded with the following weight ratio: 73 wt% of AM, 8 wt% of Csp and 19 wt% of

PVdF-HFP for Mo6S8 and LFP and 60 wt% of AM, 20 wt% of Csp and 20 wt% of PVdF-HFP

for NMC622 and LTP. NMC622 and LTP are mixed with Csp for 20 min using the Spex miller

with a ball to powder weight ratio of 11 for LTP and 8 for NMC622 prior to be mixed with

PVdF-HFP. Targeted loadings are summarized in Table M.M. 1.

Materials & Methods 209

Table M.M. 1 Summary of the practical capacity and targeted loadings considered in the study and experimental ones used.

Active material Practical capacity (mA.h/g)

Targeted loading (mg/cm²)

Experimental loading (mg/cm²)

Mo6S8 128 7 5.10 ± 1.12

LFP 172 21 16.07 ± 1.60

LTP 138 9.6 10.76 ± 1.19

NMC622 170 7 7.60 ± 0.67

The mixture is then dissolved in acetone with in the proportion of

120 mgtotal mass of powder/mLacetone and sonicated during 30 min. DBP (Dibutylphtalate, 99 %

Sigma Aldrich) is added as plasticizer and the slurry heated at 50 °C for one hour under

stirring. Then, the slurry is casted in a petri-dish and left to dry at least one hour (see

Figure M.M. 1a).

Figure M.M. 1 (a) Photography of a Mo6S8 composite Bellcore electrode casted in a petri-dish. (b) Distribution of Mo6S8 (blue) and LFP (red) electrode loading. Lines corresponds to normal distribution.

Electrodes are washed 3 times for 30 min in diethyl ether (99 %min, Alfa Aesar) to

create porosity by removing DBP and dried at 80 °C under vacuum overnight. Finally,

electrodes are punched with a 0.5 inch diameter. Loading distribution is shown in Figure

M.M. 1b. for the ≈ 150 first electrodes, demonstrating good reproducibility. Despite the

differences between experimental loadings and targeted ones: ≈ 25 % lower for Mo6S8

and LFP and ≈ 10 % higher for NMC622 and LTP, the balancing between the electrodes is

maintained to 4.2 for LFP: Mo6S8 and 1.15 for LTP:NMC622.

1.2.2 Alternative process of electrode fabrication

For the overcapacitive carbon YP50 counter electrodes, self-standing PTFE electrodes

are prepared by mixing YP50 and PTFE (Polytetrafluoroethylene, 60 wt% dispersion in

water) at a mass ratio of 9:1 in isopropanol. Isopropanol volume is added gradually until

an appropriate slurry is obtained. Then, the slurry is laminated several times following a

210 Materials & Methods

puff pastry technique to obtain films of loadings around 20 mg/cm². Finally, electrodes

are dried at 80°C under vacuum.

NMC622 electrochemical signature in LP30 is obtained using the powder mixture of

NMC622 and Csp at a weight ratio of 8:2.

1.3.1 Aqueous electrolyte

Aqueous superconcentrated electrolytes mainly rely on the use of Lithium

bis(trifluoromethanesulfonyl)imide (LiTFSI, LiN(SO2CF3)2) which was obtained from

Solvay or Solvionic and used as received. Lithium bis(pentafluoroethanesulfonyl)imide

(LiBETI, LiN(SO2CF2CF3)2) was purchased from TCI Chemicals and used as received to

make Water-in-bisalt (WiBS) electrolyte.

20 m, 15 m, 10 m, 7 m, 5 m, 3m and 1 m (mol/kg) of LiTFSI or 20 m LiTFSI : 8 m LiBETI

electrolyte solutions were prepared by mixing LiTFSI and LiBETI salts with Milli-Q

ultrapure water following Equation M.M. 1.

Equation M.M. 1:

𝑚𝑤𝑎𝑡𝑒𝑟 = 𝑚𝑠𝑎𝑙𝑡

𝐶𝑚𝑜𝑙/𝑘𝑔 ∙ 𝑀

with mwater, the mass of water to be added (kg); msalt, the mass of salt (g); Cmol/kg, the

expected concentration (mol/kg) and M, the molar mass of the salt (g/mol).

Table M.M. 2 summarizes some physico-chemical properties for the two main

electrolyte used in this study: 20 m LiTFSI and 20 m LiTFSI : 8 m LiBETI.

Table M.M. 2 Electrolyte properties and amount of water available in 20 m LiTSFI or 20 m LiTSFI : 8 m LiBETI as function of the electrolyte volume considered.

Electrolyte Velectrolyte

[µL] 𝜌

[g/mL] m [g]

𝜔𝐻2𝑂 𝜔𝐿𝑖𝑇𝐹𝑆𝐼 𝑀𝐻2𝑂

[g/mol] 𝑛𝑤𝑎𝑡𝑒𝑟 [mmol]

20 m LiTFSI

200 1.696 0.339 0.148 0.852 18

2.79

20 m LiTFSI

250 1.696 0.424 3.49

Materials & Methods 211

with Velectrolyte, volume of 20 m LiTFSI electrolyte/cell (µL); ρ, density in (g/mL)

measured by densimeter; m, the mass of electrolyte (g); ωH2O, water mass fraction;

ωLiTFSI, LiTFSI mass fraction; MH2O, molar mass of water (g/mol); nwater, the total amount

of water available (mol).

The amount of water in these electrolytes can be calculated according to Equation

M.M. 2.

Equation M.M. 2:

𝑛𝑤𝑎𝑡𝑒𝑟 (𝑚𝑜𝑙) = 𝜔𝐻2𝑂 ∙ 𝑉𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑦𝑡𝑒 ∙ 𝜌

𝑀𝐻2𝑂

1.3.2 Organic electrolyte

Pure organic solvents EC (Ethylene carbonate), DMC (Dimethyl carbonate), 1,3-

dioxolane (DOL), 1,2-dimethoxyethane (DME) were purchased from Sigma Aldrich.

Fluoroethylene carbonate (FEC) was purchased from TCI Chemicals. EC:DMC (1:1 vol%)

and DOL:DME (1:1 vol%) were prepared by mixing the appropriate amount of solvents.

To prepare the electrolytes (7M LiFSI in FEC, 1M LiTFSI : 2M LiFSI in DOL:DME + 3%

LiNO3), salts were dried in a vacuum oven at 110 °C prior to mixing. As-prepared

electrolytes were dried over molecular sieves before use.

To perform the calibration for the fluoride ion selective electrode in organic

electrolyte, EC:DMC (1:1 %vol) from Dodochem was used. LP30 (1 mol/L LiPF6 in EC:DMC

(1:1 %vol)) used for cycling experiments was purchased from Dodochem. All pure

solvents were dried over molecular sieve prior to be used for solubility measurements.

1.4.1 LiF coating

LiF coating of approximately 30 nm thickness were deposited onto metallic Li

following by reacting NF3 with metallic Li (NF3 + 3 Li → 3 LiF + ½ N2) at 175 °C for

one hour, as described in He et al.249. Current collectors were made of stainless steel

(SS). SS/Li/LiF samples were 1.27 cm in diameter for E-SEM, SEM and XRD experiments.

SS/Li/LiF samples were 0.9 cm in diameter for GC-TCD tests.

20 m LiTFSI : 8 m LiBETI

1.78358 0.446 0.10 0.9 2.48

212 Materials & Methods

1.4.2 Al2O3 coating

The Al2O3 layers were deposited on the top of metallic Li samples using an atomic

layer deposition (ALD) Picosun R-200 Advanced reactor which is a hot wall, flow through

type reaction chamber operating in thermal ALD mode using O3 (ozone) deposition

process. The AC series (AC-bench 2025) ozone generator offers state-of-the-art silent

corona discharge ozone generating technology based on O2 (99.5 %) / N2 (0.5 %) mixing

gas. The deposition temperature range was set to 140 °C, i.e. at a threshold level well

below the metallic Li melting temperature (≈c 180 °C). Trimethylaluminium (TMA)

served as precursor (tTMA = 0.1 s) and O3 (tO3 = 0.6 s) as reactant (second precursor) to

achieve the deposition of homogenous and conformal Al2O3 layers (2 to 10 nm-thick).

Electrochemical characterizations

To perform electrochemical characterization, two set ups were used, either

Swagelok® type cells or coin cells, as described in Figure M.M. 2 and Figure M.M. 3,

respectively. Swagelok devices enable to easily recover the battery assembly

(electrode/active material, separator etc.). However, the mechanical pressure applied

to the cell is hardly reproducible due to the manual pressing of the plungers, which may

influence the electrochemical behavior of the Swagelok cells and hamper the

comparison from one cell to another. On the contrary, due to the sealing step using a

crimping machine, the pressure applied is constant for coin cells (0.8 T) and reproducible

between cells. Thus, coin cell 2032 were preferred for concentration, long-term cycling

and self-discharge experiments in 2-electrode set-up.

In both set-ups, the battery relies on the same assembly. Self-standing negative and

positive electrodes of 1.27 cm² surfaces were used. Two Whatman glass fibers are used

as separators and soaked with electrolyte. On one side, the electrode is directly placed

on the plunger or the coin cell casing. On the other side, a disk of stainless steel is placed

between the electrode and plunger/casing separated by a spring.

Materials & Methods 213

Figure M.M. 2 Swagelok photography and corresponding scheme of the battery assembly.

Figure M.M. 3 Coin cell photography and corresponding scheme of the battery assembly.

The three-electrode cells used in this study are based on Swagelok 3-way T-

connector. The third electrode is inserted by the side opening connection, as illustrated

in Figure M.M. 4 and in Figure M.M. 5. Two types of three-electrode design were used,

either PFA-based or stainless steel-based ones.

Li+ insertion/de-insertion reversibility in active material (see Chapter 2) was tested

using a three-electrode PFA-based Swagelok cell (see Figure M.M. 4) with two glassy

carbon rods as current collectors. Self-standing composite active material electrode

were used as working electrode and self-standing YP50 electrode as counter electrode.

An Ag/AgCl reference electrode was used as reference electrode. Two Whatman glass

fiber separators were used. Cyclic voltammetry was performed at 1 mV/s.

214 Materials & Methods

SEI stability during self-discharge protocol determined by electrochemical impedance

spectroscopy (EIS) was assessed using a stainless steel 3-electrode Swagelok (see

Chapter 3). Mo6S8 composite electrode was used as counter electrode and LFP

composite electrode as working electrode. The reference electrode was based on a ring

of Li0.5FePO4 deposited onto a stainless steel gauze (see picture in Figure M.M. 5b). Two

glass fibers separators were used between the negative and the reference electrode and

two glass fibers separators were placed between the reference electrode and the

positive electrode. 200 µL of electrolyte were injected in the cell. After a 10 cycles pre-

cycling step at 1C, 0.5C cycling steps with a 20 h OCV resting period after the 2nd charge

were carried out. Potentio-electrochemical impedance spectroscopy (PEIS) was

performed between 1 MHz and 50 mHz with an amplitude of 20 mV.

Li0.5FePO4 was obtained by electrochemical partial delithiation. Around 500 mg of LFP

was used in a big Swagelok cell with a foil of scratched metallic Li as negative. Two

Whatman glass fibers were soaked with LP30 electrolyte. LFP powder and glass fibers

separators were separated with a celgard separator, enabling the recovery of the

powder. The cell was cycled at C/40 until reaching half lithiated Li0.5FePO4 state

(assessed by time limitation). Once the partially delithiated LFP was recovered, the

powder was washed 3 times in DMC to remove electrolyte traces, and dried under

vacuum at 80 °C overnight. Bellcore composite electrode were then made.

Figure M.M. 4 (a) Three-electrode PFA Swagelok photography and the corresponding scheme of the battery assembly. (b) Photography of the third electrode.

Materials & Methods 215

Figure M.M. 5 (a) Three-electrode stainless steel Swagelok photography and the corresponding scheme of the battery assembly. (b) Photography of the third electrode.

Operando gas monitoring during cycling (see Chapter 2) was performed using

homemade pressure cell previously developed in the lab,265,266 as reported in Figure

M.M. 6. This cell consists of Swagelok cell in which the battery assembly is placed, a gas

reservoir of ≈ 10 mL separates the battery assembly from the pressure sensor (on the

top of the cell) directly connected and controlled by the potentiostat. The 10 mL gas

reservoir was chosen to obtain a good signal to noise ratio, as pressure increase from

gas production does not exceed 100 mbar. Moreover, to maximize the pressure sensed

by the gauge, the electrode at which gas is produced is placed on the top of the battery

assembly, facing the pressure sensor. All gas-monitoring experiments were performed

in a temperature-controlled oven. Besides, a 10-hours resting period was used before

starting the electrochemical protocol to enable the pressure to be stable.

216 Materials & Methods

Figure M.M. 6 Pressure cell photography and corresponding scheme of the cell assembly.

Furthermore, gas identification (see Chapter 2) is performed by connecting a

homemade battery cell to a mass spectrometer (MS), as detailed in Ref267,268. The MS is

an ExQ gas analysis system (Hiden Analytical, USA) composed of a HAL (Hiden Analytical)

series quadrupole mass spectrometer, an ultra-high vacuum (UHV) mass spectrometer

vacuum chamber, a vacuum pumping system and a QIC series capillary inlet. The whole

system is controlled by a MASsoft 7 professional software. A photography of the cell is

shown in Figure M.M. 7a. After assembly, the online electrochemical mass

spectrometry (OEMS) cell is transferred in a temperature-controlled oven (IPP260,

Memmert, see Figure M.M. 7b). Then, the gas line and the cell are flushed and

Materials & Methods 217

pressurized to ≈ 1.3 bar with pure argon to avoid any contamination from ambient air

or glove box atmosphere. Prior to perform the galvanostatic cycling, the OEMS cell valve

connected to the mass spectrometer capillary is opened and the cell is held at open

circuit voltage for ≈ 5 hours to reach a gas-liquid equilibrium inside the cell, and

therefore to obtain a stable baseline value for all the partial pressure signals. During the

OEMS measurements, the internal cell pressure was measured by the pressure sensor

fixed on the gas line and the produced gaseous species were continuously sampled from

the cell head space to the mass spectrometer via a thin capillary (1/16″ diameter) at a

leak rate of 12.5 µL/min. The partial pressures are eventually determined for each gas

based on their mass to charge ratio (m/z). Indeed, after ionization in the ionization

source of the MS, separation in the mass analyzer and further detection in the ion

detector, the partial pressures at m/z = 2, m/z= 32 and m/z = 44 were used to determine

the evolution of H2, O2 and CO2.

Figure M.M. 7 (a) Photography and scheme of the battery assembly used to perform online electrochemical mass spectrometry (OEMS). (b) Photography of the set-up used in th temperature-controlled oven.

All battery cells were assembled in an Ar-filled glovebox (MBRAUN). Electrolytes were

saturated with argon prior to any experiment and to get entered into the glove-box to

avoid oxygen contamination. Mo6S8/LFP (1:4) full cells were assembled using stainless

steel as current collectors. LTP/NMC622 (1.1:1) full cells were assembled using stainless

steel current collector for the negative electrode (LTP) and titanium for the positive one

(NMC622). Room temperature electrochemical tests were performed using a BCS-805

potentiostat (Bio-Logic). 55 °C galvanostatic cycling, pressure cells (25 °C) and OEMS (25

°C, 55 °C) experiments were performed in a temperature-controlled oven using a MPG2

218 Materials & Methods

multichannel potentiostat (Bio-Logic). C-rate was set as 1C being equal to one Li+ cation

inserted in one hour.

Figure M.M. 8 (a) Scheme of the electrochemical cell used for RDE measurements. (b) Photography of the set-up and (c) Scheme of the electrochemical cell used for electrochemical stability window (ESW) determination and SEI stability.

Figure M.M. 8 illustrates the three-electrode set-up using an electrochemical glass

cell for electrochemical characterizations. A jacked cell connected with a chiller was used

for temperature-dependent measurements. Working electrode was mirror polished

prior to any electrochemical measurements with a Presi® polishing machine using

alumina solution of 6 µm, 0.3 µm and 0.04 µm solutions were used. Clean working

electrodes were used for each measurement.

The activation energy for the hydrogen evolution reaction (HER) measurements,

shown in Chapter 3, were performed on a biologic VSP potentiostat with a glassy carbon

electrode mounted on a rotating disk electrode setup (RDE, PINE Inc, US.). Leakless

electrode (Edaq ET069) and Pt wire were used as reference and counter electrodes,

respectively. The leakless reference electrode is based on a Ag/AgCl reference electrode

separated from the solution by a conductive junction. Temperature was set to 25 °C,

35 °C, 45 °C and 55 °C. 4 mL of 20 m LiTFSI were added in the glass cell and fresh

electrolyte was used for each temperature to avoid electrolyte crystallization due to

repeated water reduction and degassing. The leakless reference electrode was

calibrated vs a reversible hydrogen electrode (RHE) as function of temperature before

each experiment. For that, after bubbling the solution under Argon flow for 5 min to

remove oxygen dissolved in the electrolyte, the cell was left at OCV for 50 min to ensure

homogeneous temperature. The solution was then bubbled under H2 flow for 5 min and

Materials & Methods 219

the potential of the leakless electrode measured against that of Pt. For each

temperature, rotating speed was set to 400, 600, 900 and 1600 tr/min. Cyclic

voltammetry experiments were performed at 50 mV/s from 0 V vs Ref to -1.2 V vs Ref.

Prior to perform cyclic voltammetry, ohmic drop determination was assessed at 200 kHz.

Table M.M. 3 summarizes the potential of the leakless reference electrode vs the RHE

and the ohmic drop determination found experimentally.

Table M.M. 3 Calibration parameters for the determination of HER activation energy on Pt in 20 m LiTFSI by electrochemical cell.

Temperature (°C) Eref vs RHE (mV) Ohmic drop (Ω)

25 237 77

35 243 50

45 268 48

55 235 46

Data analysis was performed by rescaling the potential toward RHE and removing

Ohmic drop contribution. Current density at an overpotential (η) of -600 mV was

determined. Only the forward current was used in the study.

Electrochemical stability window (ESW) (see Chapter 2) and SEI stability were

determined by cyclic voltammetry (see Chapter 3) experiments performed in a 3-

electrode electrochemical glass cell described in Figure M.M. 8b. The ESW was

determined by performing CV using current collector materials (stainless steel or

titanium) as working electrode. Pt wire was use as counter electrode and Ag/AgCl

reference was used as reference electrode. The SEI stability was studied using a PTFE

embedded glassy carbon disc (4 mm diameter, Pine Research Instrumentation) as

working electrode and Pt wire as counter electrode. Silver wire was used as pseudo-

reference. Cyclic voltammetry experiments were performed on a VMP3 potentiostat

(Bio-Logic). 100 mV/s scan rate was applied for ESW determination and 50 mV/s for SEI

stability. All potentials were converted vs Li+/Li scale.

Physico-chemical characterizations

The densities of solutions containing a precisely known amount of salt (msalt) and

water (mwater) were measured with an electronic densitometer (Anton Paar, DMA 35

Basic).

220 Materials & Methods

The differential scanning calorimetry (DSC) measurements were done using a

NETZSCH DSC 204F1 instrument. Calibration was done measuring samples prepared by

adding two droplets of LiTFSI-based electrolyte on a 5 mm-diameter glass fiber

separators (Whatman). Prior to any experiment, electrolytes are bubbled 15 min under

Ar and separators are kept in the glovebox. The addition of electrolyte and the transfer

in the aluminum pan were done in ambient atmosphere as fast as possible to reduce

exposure to air and contamination by water traces. Unlike in previous reports, no

mesocarbon microbeads (MCMB) were added in the calibration for promoting

crystallization and avoiding supercooling126. This choice was made to remain close to the

experimental conditions for which coin cells are assembled with no MCMB.

The concentrations of the electrolyte used for calibration were calculated thanks to

Equation M.M. 1 to be 19.803 m, 20.591 m, 20.910 m, 21.992 m and 22.9983 m. This

small range of concentration for calibration was chosen remain in a concentration region

in which the liquidus line can be consider as linear (as shown in Chapter 1, Figure 1.17).

For electrolyte samples analyzed after cycling, Mo6S8/LFP coin cells were cycled

during 15, 50 and 100 cycles at 0.5 C using 60 µL of 20 m LiTFSI electrolyte. Two glass

fiber separators were used. Coin cells samples were opened in glovebox atmosphere

and 5 mm diameter samples were punched in the middle of the two separators,

assuming that the electrolyte concentration is homogeneously distributed. Once

punched, the samples were transfer in the Al pan that is sealed outside the glovebox.

For the DSC measurements, the following protocol was applied. The temperature was

first set to increase from 25 °C to 35 °C to ensure liquid state of the electrolyte, followed

by a cool down step to - 60° °C and a heating step up to 60 °C, all at 2 °C/min under

nitrogen cooling.

3.3.1 Calibration of the ISE

To determine the solubility limit of fluoride in both aqueous and organic-based

solution, a fluoride ISE from Hach Lang (Intellical, ISEF121) was used. Prior to any

solubility measurements, the electrode was calibrated using the protocols described

below.

Materials & Methods 221

Figure M.M. 9 Calibration curve of the fluoride ion selective electrode (ISE) obtained for organic (orange line) and aqueous (blue line) standard.

3.3.1.1 Aqueous standard preparation

Standard aqueous solutions (25 mL minimum) were prepared by dissolving sodium

fluoride (NaF, Alfa Aesar) in pure mQ-water at a concentration of 1000 mg/L. From this

solution, solutions of concentration of 100 mg/L; 10 mg/L; 1 mg/L; 0.1 mg/L; 0.01 mg/L

were prepared by dilution. Measurements for high concentration levels (≥10 mg/L) were

conducted by adding one pillow of TISAB (ionic strength buffer -succinic acid- TISAB,

Hach) to 25 mL of NaF solution. Low-concentration measurements (from 1 mg/L and

below) were conducted according to the following protocol: a TISAB pillow was added

to 25 mL of milli-Q ultrapure water and 5 mL of the as-prepared solution was added to

the NaF solution. Solutions were stirred and [F-] was measured.

3.3.1.2 Organic standard preparation

Organic standard solutions were prepared by dilution of 1 mol/L of TBAF in THF

solution (tetrabutylammonium fluoride solution, 1.0 M in tetrahydrofuran, Sigma

Aldrich) in EC:DMC (Dodochem, 1:1 vol%) at an initial concentration of fluoride of 1900

mg/L. By continuous dilution of a factor 11, standards with lower concentration were

prepared adding 300 µL of the organic standard to 3 mL of EC:DMC solution. The

222 Materials & Methods

following standard concentrations were prepared: 157 mg/L; 14 mg/L; 1.30 mg/L; 0.12

mg/L and 0.011 mg/L. These organic standards were then diluted by adding 2 mL in a 20

mL TISAB:H2O aqueous solution (1 TISAB pillow for 20 mL mQ-water) to perform the

experiments. Solutions were stirred and [F-] was measured.

The potential and the temperature of the standard solutions were recorded by the F-

ISE under stirring in a plastic beaker. The calibration curve obtained in both aqueous and

organic media are presented in Figure M.M. 9. A slope of -55.8 mV/decade is obtained

at 23 °C in aqueous standards while slope of -53.8 mV/decade is found in organic

standard at 23 °C. The theoretical slope is -58.7 mV/decade, and both calibrations were

considered to be sufficiently accurate to perform LiF solubility measurements.

3.3.2 Solubility limit protocol and measurement

Commercial lithium fluoride (LiF powder, 300 mesh, Sigma Aldrich) was used in

solubility tests. Saturated solutions were prepared as follows: excess amount of LiF were

added to the solution. The solution was (i) let to stir overnight (or more) at constant

temperature and (ii) let to rest before the formation of a precipitate could be visually

observed. The solution was then centrifuged for 10 min at 6000 tr/min and the

supernatant filtered using a PTFE or a polypropylene 0.2 µm pores filter. Solubility limit

measurements were performed at the thermodynamic equilibrium. Therefore, we do

not expect the solubility measurements to be dependent on the particulate size neither

on their morphologies.

One may note that in the study by Jones et al.242,269,270 solutions were not filtered but

the LiF solubility was measured from the supernatant after 24 h of decantation. This

preparation led to a colloidal solution in which ion aggregation may be present. On the

contrary, Tasaki et al.243 used a 2 µm pores filter to avoid the presence of any

undissolved salt, and thus to obtain non-colloidal solution. As explained by Jones et

al.242,269,270, measuring colloidal solutions allows for the determination of the “real”

solubility, as it takes into consideration all species in solution, i.e. from solvated ions to

aggregates. Hence, their solubility value is said to be closer to the salt species behavior

in the electrolyte. However, in our case, measurements in colloidal solution gave poorly

reproducible results as the amounts of aggregates and remaining undissolved salt could

vary depending on the supernatant collected. Therefore, we decided to measure non-

colloidal solubility by filtering the supernatant with a 0.2 µm filter. Using our protocol,

the LiF solubility limit measured in pure water (0.93 g / L at 23 °C) is close from the 1.11

g / L at 25 °C reported in the literature without filtration242, thus validating our

Materials & Methods 223

measurement protocol. Moreover, bearing in mind that superconcentrated aqueous

electrolytes were reported to be acidic,271 our solubility measurements take into

account this effect and its potential impact on the LiF solubility. Indeed, the addition of

the ionic strength buffer and dilution protocol lead to pH value within the range of 5.5

to 6, as estimated with a glass pH electrode.

Regarding the samples measured at 55°C, all the laboratory equipment (pipettes,

centrifuge tube, filter etc.) were heated at 55°C to avoid temperature artefacts. 2 mL of

the saturated solutions were then diluted in 20 mL of milli-Q ultrapure water to prepare

a diluted LiF-saturated electrolyte. Once the dilution was performed, the LiF

concentration in solution was found far from the solubility limit. We therefore

concluded that temperature control was unnecessary.

Three measurements were performed for each sample tested, except for

20 m LiTFSI : 8 m LiBETI. The mean value and standard deviation (3σ) are shown.

3.3.3 Fluoride concentration in LiF layer

Theoretical concentration of fluoride in LiF layer was calculated as following.

Considering a LiF layer of 30 nm in thickness (as determined by SEM observation), 1.27

cm in diameter and considering the density of LiF to be 2.64 g/cm3, the mass of LiF is

expected to be 1.68 .10-5 g. For a 10 nm thick layer, the mass of LiF is expected to be

3.34 .10-6 g. The mass of LiF divided by the electrolyte volume in which it is dissolved

gives the theoretical concentration ([𝐹𝑡ℎ𝑒𝑜𝑟𝑒𝑡𝑖𝑐𝑎𝑙− ]).

Experimental fluoride concentration determination of Li/LiF sample was performed

as follow: a Li/LiF sample is dissolved in a certain amount (125 or 150 mL) of TISAB:H2O

buffer solution until pH is measured to be in the range 5.5 to 6. Then, [F-] is measured

and compared to the [F-] expected when considering complete dissolution of the entire

30 nm thickness.

Relative deviation is calculated following Equation M.M. 3.

224 Materials & Methods

Equation M.M. 3:

𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 = [𝐹𝑡ℎ𝑒𝑜𝑟𝑒𝑡𝑖𝑐𝑎𝑙

− ] − [ 𝐹𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑎𝑙− ]

[𝐹𝑡ℎ𝑒𝑜𝑟𝑒𝑡𝑖𝑐𝑎𝑙− ]

𝑥 100

1.1 SEM and E-SEM

3.3.4 SEM

Scanning electron microscopy (SEM) images of the Li/LiF samples were taken with a

Zeiss Merlin scanning electron microscope at an accelerating voltage of 1 kV. Samples

were transferred to the SEM without exposure to air via a transfer vessel (Semilab Inc.)

built for the Zeiss SEM airlock. The average and standard deviation of the LiF layer

thickness were determined by measurements at five random positions on the edge of

LiF layer in the cross-section view with tilt angle correction.

3.3.5 E-SEM

Li, Li/LiF and Li/Al2O3 samples of 1.27 cm, 1.27 cm and 0.7 cm in diameter,

respectively, are affixed to sample holders using Cu tape in an Ar-filled glovebox (< 0.1

ppm O2, < 0.1 ppm H2O). The samples are transferred from the Ar-filled glovebox to an

environmental SEM (ESEM Quattro S by Thermo Fisher Scientific) with minimal exposure

to air and loaded into the chamber under high vacuum. The SEM is used in high-vacuum

(0 % relative humidity, RH) and environmental (5, 30, 45, 60, 75 and 90 % RH, see Figure

M.M. 10) mode for SEM imaging at 20 °C with a GSED detector in secondary electron

(topography) mode. After recording a reference image at 0 % RH in high-vacuum mode,

the SEM is operated in E-SEM mode at 5 % RH. The samples are then exposed to 15 %

RH for 2 minutes and imaged at 5 % RH; alternating between reaction (high RH for 2

minutes) and imaging (always 5 % RH) conditions and repeated for 30, 45, 60, 75 and 90

% RH (see depiction of samples exposed to 30, 60 and 90 % RH and imaged at 5 % RH in

Chapter 2). The SEM images are taken at 1.5/1.8 nA and 10/15 kV at a working distance

of 8 mm, and a resolution of 1536 x 1024 pixel with a horizontal field width of 207 µm.

Materials & Methods 225

Figure M.M. 10 Working principle of increasing humidity in E-SEM.

Inside an Ar-filled glovebox (< 0.1 ppm O2, < 0.1 ppm H2O), a gas-tight cell made from

chemically-inert polyetheretherketone (PEEK) was assembled with Li/LiF or Li/Al2O3

samples inside. The cell has a 5.5 mL gas headspace, inlet and outlet valves, and septa

through which liquids can be injected and gas from the headspace can be extracted with

gas-tight syringe. The gas-tight cell was then taken outside of the glovebox, connected

to an Ar tank (R300, Airgas) and purged for 5 min at 100 mgAr/min. The cell was then

filled with Ar to a pressure of 30 psi. 250 µL of electrolyte (20 m LiTFSI or 20 m LiTFSI : 8

m LiBETI) was then injected into the cell through a septum with a gas-tight syringe. Every

15 min, a 2.5 mL gas sample was collected with a gas-tight syringe and the remaining Ar

headspace was purged with fresh Ar at 100 mgAr/min for 1 minute and the pressure set

back to 30 psi. The gas samples were then injected into an Agilent 7890B gas

chromatography instrument equipped with thermal conductivity (TCD) and flame

ionization (FID) detectors for gas analysis, calibrated using 15 ppm and 1 vol% gas

standards in N2 (Supelco). The TCD detector was used for H2 and CO2 quantification, and

the FID detector was used for CH4, C2H2, C2H4, C2H6 and CO quantification.

The pre-soaking of the LiF-coated samples was done by pipetting 50 µL (or 2 mL for

pure FEC) of the organic electrolytes on LiF-coated Li samples inside a glovebox, which

was left to react for 1 h, and subsequently dried under antechamber vacuum for 1 h

before exposure to aqueous superconcentrated electrolytes. As the exposure to organic

FSI--based electrolytes lasts one hour followed by a one-hour drying step, FSI- is not

directly exposed to water. If so, the time during which (two hours) the remaining traces

226 Materials & Methods

of FSI- anions may be exposed to either WiSE or WiBS is far shorter than the time

necessary to detect a significant FSI- hydrolysis 27.

X-ray photoelectron spectroscopy (XPS) was conducted on a PHI VersaProbe II X-ray

Photoelectron Spectrometer. Samples were transferred to XPS in an air-sensitive

transfer vessel to minimize exposure to air. Binding energies were calibrated by the

adventitious carbon peak at 284.8 eV.

Data treatment

4.1.1 Contribution to direct HER in Mo6S8/LFP cell irreversible capacity

Calculations of the amount of gas evolved during cycling can be done using two

different experimental set-ups. The first one is using pressure cell data while the second

one is using OEMS data. Assumptions made are the following:

The HER mechanism considered during these calculations is the direct two electrons

HER (see Equation M.M. 4).

Equation M.M. 4: Direct HER

2𝐻2𝑂 + 2𝑒− → 2𝐻𝑂− + 𝐻2

We estimated that the pressure evolution occurs only during second insertion

plateau during charge, as observed in Figure M.M. 11.

Materials & Methods 227

Figure M.M. 11 Illustration of operando gas monitoring during pressure cell experiment (based on Mo6S8/LFP in 20 m LiTFSI). Black line represents potential as function of time. Red line is the pressure as function of time. Experiment performed at 0.10C, 25 °C.

Pressure cell data are smoothed according to adjacent-averaging method on 30

points in Originlab software version 8. Figure M.M. 11 shows both the pressure

evolution and the potential as a function of time. From the ideal gas law (see Equation

M.M. 5), variation of pressure can be linked to the amount of mole of gas produced. All

pressure analysis were performed by taking as minimal and maximal pressure values the

pressure values corresponding to the green arrows as shown in Figure M.M. 11. The

relation between the amount of gas evolved and the capacity is given by the Faraday’s

law (see Equation M.M. 6). In addition, the irreversible capacity (see Equation M.M. 7)

is determined by cycling performances analysis. Then, the irreversible capacity due to

HER is rationalized by the total irreversible capacity, as described in Equation M.M. 8.

Equation M.M. 5: Ideal gas law

𝑛𝑔𝑎𝑠(𝑚𝑜𝑙) = 𝛥𝑃 ∙ 𝑉

𝑅 ∙ 𝑇 (𝑚𝑜𝑙)

with 𝛥𝑃 the pressure increase during cycling (Pa); V the volume of the pressure cell

(m3) (here, 9.17 mL), R the molar gas constant (J/mol/K) and T the temperature (K).

Equation M.M. 6: Faraday's law

228 Materials & Methods

𝑄𝐻𝐸𝑅(C) = 2𝑛𝑔𝑎𝑠𝐹

with ngas (mol) defined by Equation M.M. 5, F the Faraday constant (96500 C/mol), 2

the number of electron exchanged.

Equation M.M. 7: Irreversible capacity calculation

𝑄𝑖𝑟𝑟(𝐶) = 𝑄𝑐ℎ𝑎𝑟𝑔𝑒 − 𝑄𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒

with Qcharge, the charge capacity of a cell, Qdischarge the discharge capacity.

Equation M.M. 8: 𝑄𝐻𝐸𝑅𝑄𝑖𝑟𝑟

4.1.2 Indirect HER in Mo6S8/LFP

The indirect HER mechanism can be expressed by the Equation M.M. 9, as being the

sum of the following two half reactions:

𝐿𝑖4𝑀𝑜6𝑆8 → 𝐿𝑖4−𝑥𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐿𝑖+ + 𝑥 ∙ 𝑒−

+

𝑥 ∙ 𝐻2𝑂 + 𝑥 ∙ 𝑒− → 𝑥 ∙ 𝐻𝑂− + 𝑥

2∙ 𝐻2

___________________________________________________

Equation M.M. 9: Indirect HER

𝐿𝑖4𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐻2𝑂 → 𝐿𝑖4−𝑥𝑀𝑜6𝑆8 + 𝑥 ∙ 𝐿𝑖+ + 𝑥 ∙ 𝐻𝑂− + 𝑥

2∙ 𝐻2

Figure M.M. 12 Illustration of pressure cell experiment using self-discharge protocol. Black line represents the potential as function of time. Pressure evolution as function of

Materials & Methods 229

time is shown by the red line. (a) Repeated self-discharge period. (b) Zoom on open circuit voltage (OCV) period. Experiments performed at 0.5C, 25 °C.

The amount of mole of hydrogen produced during self-discharge can be calculated

from the Equation M.M. 5 and related to a capacity loss by HER (𝑄𝐻𝐸𝑅 𝑑𝑢𝑒 𝑡𝑜 𝑟𝑒𝑠𝑡𝑖𝑛𝑔)

thanks to Equation M.M. 6. The capacity loss by parasitic reaction (𝑄𝑙𝑜𝑠𝑡 𝑏𝑦 𝐻𝐸𝑅) during

resting period was calculated according to Equation M.M. 10 from cycling data. Then,

the proportion of HER in the total capacity loss during self-discharge is calculated thanks

to Equation M.M. 11.

Equation M.M. 10:

𝑄𝑙𝑜𝑠𝑡 𝑏𝑦 𝐻𝐸𝑅 = 𝑄𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒 𝑛 − 𝑄𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒

𝑛+1 (𝐶)

Equation M.M. 11:

𝑟𝑎𝑡𝑖𝑜 = 𝑄𝐻𝐸𝑅 𝑑𝑢𝑒 𝑡𝑜 𝑟𝑒𝑠𝑡𝑖𝑛𝑔

𝑄𝑙𝑜𝑠𝑡 𝑏𝑦 𝐻𝐸𝑅

Two hydrogen evolution slopes are observed during charge (see red arrows, part 1

and 2 in Figure M.M. 13). At the end of charge (part 2), one can assume that the only

reaction occurring is the HER. Therefore, the capacity corresponds to the capacity due

to HER ( Δ𝑄𝐻𝐸𝑅 𝑝𝑎𝑟𝑡 2) . Two C-rate were used: 0.15C and 0.10C. Table M.M. 4

summarizes the parameters that enable to find the conversion constant between uma

and mole of hydrogen produced during cycling, following Equation M.M. 12 and

Equation M.M. 13. Once the conversion parameter found, the total amount of hydrogen

produced during charge can be calculated and compared to the irreversible capacity,

alike what is done to analyze results from pressure cells (see Equation M.M. 6, Equation

M.M. 7 and Equation M.M. 8).

230 Materials & Methods

Figure M.M. 13 Illustration of operando gas monitoring during OEMS experiment (based on Mo6S8/LFP in 20 m LiTFSI). Black line represents potential as function of time. Red line is the hydrogen evolution as function of time. Experiments performed at 0.15C, 25 °C.

Table M.M. 4 Summary of parameters used to obtain the conversion constant between uma and moles of hydrogen produced during cycling.

C-rate

Current (mA)

Δ𝑄𝐻𝐸𝑅 𝑝𝑎𝑟𝑡 2

(𝐶)

𝑛𝐻2 𝑝𝑎𝑟𝑡 2(µmol)

𝑛𝐻2 𝑝𝑎𝑟𝑡 2(uma) 1∙10-10 Uma

to µmol

𝑛𝐻2 𝑡𝑜𝑡𝑎𝑙

(µmol)

0.15C 0.0303 0.501 2.60

4.3∙10-10 0.605 3.48

0.10C 0.0117 0.216 1.117

8.712∙10-11 0.1822 0.216

Equation M.M. 12:

Δ𝑄𝐻𝐸𝑅 𝑝𝑎𝑟𝑡 2 = 𝑖Δ(𝑡)𝑝𝑎𝑟𝑡 2 (𝐶)

Equation M.M. 13:

𝑛𝐻2 = Δ𝑄𝐻𝐸𝑅 𝑝𝑎𝑟𝑡 2

2𝐹 (𝑚𝑜𝑙)

Metallic Li consumption during GC-TCD experiments were calculated by comparing

the initial amount of Li available in the sample with the amount of hydrogen produced,

which corresponds to a molar balance of one mole of H2 produced for two moles of

Materials & Methods 231

metallic Li consumed, according to Equation M.M. 14. Table M.M. 5 lists the

experimental parameters needed to calculate the amount of hydrogen produced. All the

calculations consider hydrogen as being solely produced by direct HER. Equation M.M.

15, Equation M.M. 16 and Equation M.M. 17 detail the calculations needed to determine

the ratio of metallic Li consumed by water reduction.

2𝐻2𝑂 + 2𝑒− → 2𝐻𝑂− + 𝐻2

+

𝐿𝑖 → 𝐿𝑖+ + 𝑒−

_____________________________________________________________

Equation M.M. 14:

𝐻2𝑂 + 𝐿𝑖 → 𝐿𝑖𝑂𝐻 + 1

2𝐻2

Table M.M. 5 Experimental parameters for GC-TCD/FID calculations.

*based on SEM observation

Equation M.M. 15: Amount of mole of gas available in the cell’s headspace.

𝑛𝑐𝑒𝑙𝑙𝑔𝑎𝑠

= 𝑃𝑡𝑜𝑡 (𝑃𝑎) ∙ 𝑉𝑐𝑒𝑙𝑙 (𝑚3)

𝑅 ∙ 𝑇= 689.8 µ𝑚𝑜𝑙

Equation M.M. 16: Amount of mole of metallic Li initially available.

𝑛𝐿𝑖 𝑓𝑜𝑖𝑙 = ρ ∙ 𝜋 ∙ Ø2 ∙ e

4 ∙ M= 489 µ𝑚𝑜𝑙 (𝐿𝑖𝐹) 𝑜𝑟 90 µ𝑚𝑜𝑙 (𝐴𝑙2𝑂3)

𝑛𝐻𝑦𝑑𝑟𝑜𝑔𝑒𝑛 = 𝑛𝑐𝑒𝑙𝑙𝑔𝑎𝑠

∙ 𝑥𝐻𝑦𝑑𝑟𝑜𝑔𝑒𝑛∗ (𝑚𝑜𝑙)

Equation M.M. 17:

𝑅𝑎𝑡𝑖𝑜 𝑜𝑓 𝐿𝑖 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 = 2 ∙ 𝑛𝐻𝑦𝑑𝑟𝑜𝑔𝑒𝑛

𝑛𝐿𝑖 𝑓𝑜𝑖𝑙

*The molar ratio of hydrogen (𝑥𝐻𝑦𝑑𝑟𝑜𝑔𝑒𝑛∗ ) is obtained from the sum of the amount of

H2 measured after each 15 min interval.

Volume of the

cell 𝑉𝑐𝑒𝑙𝑙 (mL)

Total pres- sure 𝑃𝑡𝑜𝑡 (bar)

Tempe- rature T (°C)

Ideal gas

constant 𝑅

(J/mol/K)

Thick- ness of Li foil* e (µm)

Diameter of Li foil Ø (mm)

Li den- sity ρ

(g/cm3)

Molar mass of Li M

(g/mol)

5.5 3.06 20 8.314 LiF Al2O3 LiF Al2O3

0.534 6.941 100 30.5 9 7

232 Materials & Methods

Table M.M. 6 summarizes the cycling performances obtained during the pre-cycling

step at 25 °C prior to perform the temperature-dependence study. One can notice that

similar behaviors are obtained regardless the cell considered.

Table M.M. 6 Cycling performances obtained during pre-cycling step of Arrhenius study. Mean charge and discharge capacity obtained during 10 pre-cycling cycles at 25 °C for each temperature tested. Cells are based on Mo6S8/LFP full cells cycled using 20 m LiTFSI.

1.2 Impact of the electrolyte volume on cell performances: number of cycles before crystallization at 21 m LiTFSI

For the 20 m LiTFSI electrolyte with the properties described in Table M.M. 1, the

amount of mole of water in 20 m LiTFSI is defined according to Equation M.M. 18.

Protocol Number of cells

Temperature (°C) 15 25 35 45 55

HER direct by pressure

cell 1

Qchargepre−cycling at 25 °C

(mA.h/g)

103 ± 7

103 ± 7

108 ± 6

106 ± 6

Qdischargepre−cycling at 25 °C

(mA.h/g)

100 ± 4

95 ± 1.5

103 ± 3

102 ± 4

HER direct by CC-CV coin cells

2

Qchargepre−cycling at 25 °C

(mA.h/g)

117 ± 2

117 ± 2

112 ± 2

117 ± 2

Qdischargepre−cycling at 25 °C

(mA.h/g)

109 ± 1

112 ± 2

109 ± 1

115 ± 2

HER indirect by

self-discharge

on 2nd plateau

3 (except

2 at 15 °C)

Qchargepre−cycling at 25 °C

(mA.h/g)

106 ± 4

109 ± 3

108 ± 3.5

107 ± 3

109 ± 3

Qdischargepre−cycling at 25 °C

(mA.h/g)

103 ± 3

107 ± 2

105 ± 2

105 ± 2

106 ± 2

HER indirect by

self-discharge

on 1st plateau

3

Qchargepre−cycling at 25 °C

(mA.h/g)

113 ± 2

107 ± 4

118 ± 5

109 ± 4

Qdischargepre−cycling at 25 °C

(mA.h/g)

111 ± 1

103 ± 2

115 ± 4

106 ± 3

Materials & Methods 233

Equation M.M. 18:

𝑛20 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼𝑤𝑎𝑡𝑒𝑟 =

𝜌20 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼 ∙ 𝑉𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 ∙ 𝜔𝑤𝑎𝑡𝑒𝑟

𝑀𝑤𝑎𝑡𝑒𝑟 (𝑚𝑜𝑙)

with 𝜌20 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼 , the density (g/cm3); 𝑉𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 , the electrolyte volume (cm3);

𝜔𝑤𝑎𝑡𝑒𝑟 , the mass fraction of water and 𝑀𝑤𝑎𝑡𝑒𝑟, the molar mass of water (g/mol).

Fixing the amount of salt between 21 m LiTFSI and in 20 m LITFSI (thus only

considering water consumption), the amount of mole of water in 21 m LiTFSI is defined

according to Equation M.M. 19.

Equation M.M. 19:

𝑛21 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼𝑤𝑎𝑡𝑒𝑟 =

𝑚𝐿𝑖𝑇𝐹𝑆𝐼

𝐶 ∙ 𝑀𝐿𝑖𝑇𝐹𝑆𝐼 ∙ 𝑀𝑤𝑎𝑡𝑒𝑟 (𝑚𝑜𝑙)

with 𝑚𝐿𝑖𝑇𝐹𝑆𝐼, the mass of LiTFSI salt initially present in 20 m LiTFSI (g); 𝐶, the expected

concentration (here 21 m LiTFSI) (mol/kg); 𝑀𝐿𝑖𝑇𝐹𝑆𝐼, the molar mass of LiTFSI (g/mol).

One cycling protocol consists of 10 cycles performed at 1C as pre-cycling step

followed by a cycling step performed at 0.5C consisting of charge/discharge/charge

followed by 20 h of OCV and a discharge/charge/discharge. The amount of mole of water

consumed during such protocol is calculated by Equation M.M. 20, and values are

summarized in Table M.M. 7

Equation M.M. 20:

nduring protocolwater consumed = 2 ∙

𝑄𝑖𝑟𝑟𝐻𝐸𝑅

𝑧∙𝐹 (mol)

with 2, a factor corresponding to the stoichiometric coefficient between H2 and H2O

production; 𝑄𝑖𝑟𝑟𝐻𝐸𝑅, the irreversible capacity due to HER during both cycling and resting

period (C); z, the number of electron involved in the HER reaction; F, the faraday

constant (C/mol). 𝑄𝑖𝑟𝑟𝐻𝐸𝑅, the irreversible capacity due to HER is calculated assuming that

HER account for 70 % of the irreversible capacity during cycling and 100 % of the

irreversible capacity during self-discharge.

Equation M.M. 21 details the calculation made to determine the number of times the

protocol described above can be repeated before 21 m LiTFSI concentration is reached

due to water consumption.

234 Materials & Methods

Equation M.M. 21:

𝑁𝑏𝑒𝑓𝑜𝑟𝑒 21 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼𝑝𝑟𝑜𝑡𝑜𝑐𝑜𝑙 =

𝑛20 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼𝑤𝑎𝑡𝑒𝑟 − 𝑛21 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼

𝑤𝑎𝑡𝑒𝑟

𝑛𝑑𝑢𝑟𝑖𝑛𝑔 𝑝𝑟𝑜𝑡𝑜𝑐𝑜𝑙𝑤𝑎𝑡𝑒𝑟 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑

Table M.M. 7 Amount of water available in 20 m LiTFSI and 21 m LiTFSI assuming the initial same amount of LiTFSI salt in both electrolyte. Amount of water consumed per protocol (µmol). Number of protocol that can be done before electrolyte crystallization.

Velectrolyte

(µL) n20 m LiTFSIwater available

(mol) n21 m LiTFSIwater available

(mol)

nduring protocolwater consumed

(µmol) 𝑁𝑏𝑒𝑓𝑜𝑟𝑒 21 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼𝑝𝑟𝑜𝑡𝑜𝑐𝑜𝑙

9.4 0.000131221 0.000125338 ≈ 5 1.2

50 0.000697244 0.000665981 4.151 7.5

100 0.001394489 0.001331962 4.396 14.2

150 0.002091733 0.001997943 5.429 17.3

200 0.002788978 0.002663923 4.391 28.5

300 0.004183467 0.003995885 4.270 43.9

400 0.005577956 0.005327847 5.570 44.9

500 0.006972444 0.006659809 3.767 83

600 0.008366933 0.00799177 3.917 95.8

Python

Fitting data for solubility is thus obtained by using “curve_fit” function from SciPy in

Python.

LIST OF ABBREVIATIONS

236 Abbreviations

𝑎𝐹− Activity of F-

𝑎𝐿𝑖+ Activity of Li+

Å Angstrom

AC Acetate

ACN Acetonitrile

Al(OH)3 Aluminum hydroxyde

Al2O3 Aluminum oxide

ALD Atomic layer deposition

AM Active material

CCCV Constant current constant voltage

Cd Cadmium

CE Coulombic efficiency

C-rate Cycling-rate

Csp Carbon super P

CV Cyclic voltametry

DBP Dibutylphtalate

DEC Diethylcarbonate

DMC Dimethyl carbonate

DME Dimethoxyethane

DMF Dimethylformamide

DMPA 2,2-dimethoxy-2-phenylacetophenoe

DMSO Dimethyl sulfoxide

DOL 1,3-dioxolane

DSC Differential scanning calorimetry

E Potential

e Thickness

EC Ethylene carbonate

EIS Electrochemical impedance spectroscopy

EMC Ethyl methyl carbonate

EMim TFSI 1-Ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide

E-SEM Environmental-scanning electron microscopy

ESW Electrochemical stability window

F Faraday constant

[F-]lim Maximum of F- concentration

FEC Fluoroethylene carbonate

FEMC 2-fluoroethyl methyl carbonate

GC Glassy carbon

GC-TCD/FID Gas chromatography-thermal conductivity/flame ionization

detectors

GPE Gel polymer electrolyte

H2O Water

H2SO4 Sulfuric acid

HEA 2-hydroxyethyl acrylate

HER Hydrogen evolution reaction

Abbreviations 237

HFE Hughly fluorinated ether

i Current

IEA International energy agency

IL Ionic liquids

ISE Ion selective microscope

KCl Potassium chloride

KOH Potassium hydroxide

Ks Constant of dissolution of LiF

KTFSI Potassium(I) Bis(trifluoromethanesulfonyl)imide

LCO LiCoO2 Lithium cobalt oxide

LFP Lithium iron phosphate

[Li+]lim Maximum of Li+ concentration [𝐿𝑖+]𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒 Concentration of Li+ in the electrolyte

[𝐿𝑖+] Concentration of Li+

Li(H2O)4+ Hydrated lithium cluster

Li2CO3 Lithium carbonate

LiAc Lithium acetate

LiAsF6 Lithium hexafluoroarsenate

LIB Li-ion battery

LiBETI Lithium bis(pentafluoroethanesulfonyl)imide

LiFSI Lithium bis(fluorosulfonyl)imide

LiFTFSI Lithium (fluorosulfonyl)(trifluoromethylsulfonyl)imide

LiMM3411 Lithium propylsulfonate-butylsulfonate-dimethyl-amine

LiNO3 Lithium nitrate

LiOtf Lithium triflouromethanesulfonate or Lithium triflate

LiPF6 Lithium hexafluorophosphate

LiPTFSI Lithium pentafluoroethanesulfonyl)

(trifluoromethanesulfonyl)imide

LiTFSI Lithium bis(trifluoromethanesulfonyl)imide

LP30 1 mol/L LiPF6 in EC:DMC (1:1 vol%)

LSV Linear scanning voltametry

LTO Lithium titanate

M Molar mass

m/z Mass to charge ratio

ME3EtN+ Trimethylethylammonium

MPEGA poly(ethylene glycol) methyl ether acrylate

MS Mass spectrometer

msalt Mass of salt

mwater Mass of water

N/P Negative to positive capacity ratio

NCA Lithium nickel cobalt aluminum oxide

ngas Amount of mole of gas produced

𝑛20 𝑚 𝐿𝑖𝑇𝐹𝑆𝐼𝑤𝑎𝑡𝑒𝑟 Amount of mole of water in 20 m LiTFSI

238 Abbreviations

𝑛𝐻𝑦𝑑𝑟𝑜𝑔𝑒𝑛 Amount of mole of hydrogen

𝑛𝐿𝑖 𝑓𝑜𝑖𝑙 Amount of mole of metallic Li

𝑛𝑐𝑒𝑙𝑙𝑔𝑎𝑠

Amount of mole of gas in the cell

Ni-Cd Nickel-Cadmium

Ni-MH Nickel Metal Hydride

NiOOH Nickel oxide hydroxide

NMC/NCM Lithium nickel manganese cobalt oxide

Ø Diameter

OCV Open circuit voltage

OEMS Online electrochemical mass spectrometry

OER Oxygen evolution reaction

PAM Polyacrylamide

Pb Lead

PC Propylene carbonate

PEG/PEO poly(ethylene glycol)/poly(ethylene oxide)

PEGDA poly(ethylene glycol) diacrylate

PEIS Potentio-electrochemical impedance spectroscopy

PFA Perfluoroalkoxy alkane

PVdF-HFP Poly(vinylidenefluoride-hexafluoropropylene

Qcharge Charge capacity

Qdischarge Discharge capacity

QHER Capacity due to HER

Qirr Irreversible capacity 𝑄𝐻𝐸𝑅 𝑑𝑢𝑒 𝑡𝑜 𝑟𝑒𝑠𝑡𝑖𝑛𝑔 Capacity due to HER during resting

𝑄𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒 𝑛 Discharge capacity at cycle number n

𝑄𝑑𝑖𝑠𝑐ℎ𝑎𝑟𝑔𝑒𝑛+1 Discharge capacity at cycle number n+1

𝑄𝑙𝑜𝑠𝑡 𝑏𝑦 𝐻𝐸𝑅 Capacity lost by HER

R Constant of ideal gas law

RDE Rotating disk electrode

RH Relative humidity

RHE Reversible hydrogen electrode

RT Room temperature

SEI Solid electrolyte interphase

SL Sulfolane

SLI Starting lighting ignition

SOC State-of-charge

SPE Solid polymer electrolyte

SS Stainless steel

T Temperature

TBAF in THF Tetrabutylammonium fluoride solution in Tetrahydrofuran

TEA+ Tetraethylammonium

TEGDME Tetraethylene glycol dimethyl ether

TEP Triethyl phosphate

Abbreviations 239

TISAB Total ionic strength adjustment buffer

TMA Trimethylaluminium

TMP Trimethyl phosphate

TMSB Tris(trimethylsilyl)borate

UHV Ultra-High Vacuum

UPS Unit power supply

USD United States Dollar

V Volume

VRLA Valve Regulated Lead Acid

WiSE Water-in-salt electroylte

XRD X-ray diffraction

Zn(TFSI)2 Zinc(II) Bis(trifluoromethanesulfonyl)imide

ΔP Pressure increase

�̅�𝐿𝑖+𝐹− Mean ionic activity coefficient of the LiF salt

𝛾𝐹− Ionic coefficient of F-

𝛾𝐿𝑖+ Ionic activity coefficient of Li+

η Viscosity

λKα1 Wave length of Cu Kα radiation

ρ Volumetric mass

σ Conductivity

𝜔 Mass fraction

𝑥𝐻𝑦𝑑𝑟𝑜𝑔𝑒𝑛∗ Mole fraction of hydrogen

RÉSUMÉ EN FRANÇAIS

242 Résumé en français

Le remplacement des énergies fossiles par leurs pendants renouvelables est au cœur

de la transition énergétique mise en place pour limiter le changement climatique.

Cependant, les limitations d’émissions de gaz à effet de serre (tels que le dioxyde de

carbone CO2 ou le méthane CH4) peuvent entrer en conflit avec la demande croissante

en énergie. Bien que la pandémie de la Covid 19 aie contracté la demande mondiale en

énergie de 4 % en 2020, l’augmentation attendue en 2021 est de 4.1 %, selon les

prévisions de l’IEA (Agence internationale de l’Energie). Ceci est notamment due à la

croissance de la consommation dans les pays émergents et dans les économies en

développement. La mise en place de politiques publiques efficaces et d’investissements

massifs (jusqu’à quatre mille milliards de dollars par an jusqu’en 2030 comme

mentionné dans le rapport sur les perspectives énergétiques mondiales publié par l’IEA

en 2021) est nécessaire pour atteindre les objectifs discutés lors de l’Accord de Paris

pour le climat, qui vise à limiter l’augmentation de température à 2 °C, voir 1.5 °C, d’ici

la fin du siècle. Des modifications dans les modes de production d’énergie, la réduction

de la consommation d’énergie des infrastructures déjà présentes et l’amélioration de

l’efficacité énergétique peuvent être vus comme les principaux axes de développement

permettant de réussir la transition énergétique. Pour atteindre ces objectifs, le

développement de l’électrification est l’un des moyens envisagés. Cette tendance est

confirmée par les prévisions de demande en électricité qui devrait augmenter de 4.5 %

en 2021, soit cinq fois plus que la baisse de consommation de 2020. Toutefois, la

production d’électricité par les énergies renouvelables telles que le solaire ou l’éolien

est intermittente rendant primordiale le développement de dispositifs de stockage.

Parmi les dispositifs de stockage, l’utilisation de batteries permet de stocker par voie

électrochimique l’énergie produite de manière intermittente. Ces dernières reposent

sur deux électrodes, une positive et une négative dont les potentiels redox sont

différents. Afin d’assurer le transport ionique, les deux électrodes sont séparées par un

électrolyte, composé d’un solvant et d’un sel support. Le passage des électrons et donc

du courant est assuré par le circuit électrique externe. Depuis la réalisation de la

première batterie acide-plomb en 1859 par Gaston Planté, de nombreuses avancées ont

été réalisées au niveau de la chimie de la batterie, de l’ingénierie et de l’optimisation de

la cellule ainsi que du développement des usines de production, permettant d’améliorer

les performances et la fiabilité tout en réduisant les coûts. Depuis sa première

commercialisation en 1991 pour des appareils électroniques, les dernières années ont

vu l’avènement de la technologie batterie Li-ion pour de nouveaux marchés comme les

véhicules électriques. Cette technologie repose sur le mécanisme dit « rocking chair »

d’intercalation des cations Li+ dans les électrodes, et les prévisions de croissance du

Résumé en français 243

marché des batteries Li-ion (LIB) montrent que la demande va continuer d’augmenter

dans les années à venir, avec des prévisions de marché atteignant 16 mille milliards de

dollars en 2050. L’anticipation des questions de recyclage des batteries ainsi que le

développement de matériaux plus durables sont donc des enjeux cruciaux. Or, l’un des

désavantages des LIB actuellement commercialisées concerne l’électrolyte. En effet,

celui-ci repose sur l’utilisation de solvants organiques, inflammables et toxiques qui

questionnent la sécurité et la durabilité de ces systèmes. Le remplacement ces solvants

par l’eau pourrait alors apporter une solution face à ces enjeux.

Les batteries rechargeables reposant sur un électrolyte aqueux sont connues et

commercialisées depuis de nombreuses années. On peut notamment citer les batteries

de type Acide-plomb utilisées comme batterie de démarrage dans les voitures, Nickel-

Cadmium (Ni-Cd), présentes dans les premiers téléphones portables ou Nickel-Métal

Hydrure (Ni-MH) commercialisées sous format AA ou AAA et dans les premiers véhicules

hybrides électriques (tel que la Toyota Prius). Cependant, leur densité d’énergie et leur

énergie spécifique sont limitées par l’étroitesse de la fenêtre électrochimique de l’eau

(1.23 V thermodynamiquement). Au-delà des limites de cette fenêtre, les réactions

parasites d’oxydation de l’eau (oxygen evolution reaction (OER), en anglais) et de

réduction de l’eau (hydrogen evolution reaction (HER), en anglais) ont lieu au détriment

de la stabilité de l’électrolyte, comme décrit par les Equation Résumé. 1 et Equation

Résumé. 2, respectivement.

Equation Résumé. 1: Réaction d’oxydation de l’eau

2 ∙ 𝐻2𝑂 = 𝑂2 + 4 ∙ 𝐻+ + 4 ∙ 𝑒−

Equation Résumé. 2: Réaction de réduction de l’eau

2 ∙ 𝐻2𝑂 + 2 ∙ 𝑒− → 2 ∙ 𝐻𝑂− + 𝐻2 𝑜𝑢 2 ∙ 𝐻+ + 2 ∙ 𝑒− = 𝐻2

Concernant les batteries LIB aqueuses, la dégradation de l’électrolyte a lieu avant

l’intercalation des cations Li+. Contrairement aux batteries LIB commerciales, la

formation d’une interphase solide/électrolyte (Solid electrolyte interphase, en anglais

(SEI)) n’est pas observée. Or, la formation de cette dernière est à l’origine de la plus

grande densité d’énergie des batteries LIB commerciales car elle permet d’opérer au-

delà des limites de la fenêtre électrochimique de stabilité. En effet, les électrolytes

organiques classiquement utilisés, et plus précisément les solvants, sont instables à bas

potentiels (E < 1 V vs Li+/Li), i.e., aux potentiels où opèrent les électrodes négatives telles

que le graphite. La dégradation de l’électrolyte au niveau de l‘électrode négative au

cours des premiers cycles d’utilisation forme une couche passivante, isolante

244 Résumé en français

électroniquement et conductrice ionique qui prévient la dégradation supplémentaire de

l’électrolyte et assure le maintien des performances. Cette couche est généralement

issue des produits de décomposition du solvant, structurée par une couche interne riche

en composés inorganiques et une couche externe riche en composés organiques. Pour

résumer, l’absence de formation de SEI est l’un des freins au déploiement à grande

échelle des batteries LIB reposant sur un électrolyte aqueux.

En 2015, le développement d’un électrolyte aqueux superconcentrés formé de 21

mol/kg de sel de LiTFSI, c’est-à-dire, un électrolyte dont la masse et le volume de sels

sont supérieurs à ceux du solvant, a permis d’étendre la fenêtre électrochimique jusqu’à

3 V. Ces travaux sont inspirés par des études préalablement conduites sur des

électrolytes organiques superconcentrés qui ont montré l’influence positive de

l’augmentation de la concentration sur les propriétés de l’électrolyte. Dans le cas d’un

électrolyte aqueux, l’augmentation de la concentration modifie la structure de

l’électrolyte en créant deux réseaux interpénétrés : un réseau riche en anion, ici TFSI-,

et un réseau riche en eau qui maintient le transport rapide des cations Li+ malgré la

diminution du nombre de porteurs de charges dissociés. La réactivité interfaciale est

aussi impactée. Au niveau de l’électrode positive, les anions TFSI- peuplent l’interface

par un effet de double couche permettant de repousser les molécules d’eau et donc de

prévenir l’oxydation de l’eau. Au niveau de l’électrode négative, le dépôt d’une couche

inorganique passivante par la dégradation du sel atteste de la formation d’une SEI. Ces

changements de propriétés physico-chimiques et de réactivité interfaciale laissent

penser que des batteries basées sur des électrolytes aqueux pourraient concurrencer

les performances des batteries commerciales tout en améliorant leur durabilité. Ainsi,

l’objectif de cette thèse est d’étudier la viabilité de l’utilisation d’électrolyte aqueux

superconcentrés pour une application dans les LIB.

Pour cela, la mise en place d’une étude systématique a été réalisée au niveau des

limites cathodiques (à l’électrode négative) et anodique (à l’électrode positive) afin de

déterminer la stabilité de l’électrolyte basé sur 20 m LiTFSI en fonction des conditions

d’opération. Bien que l’augmentation de la concentration ait montré un impact positif

sur les performances, ces résultats peuvent être nuancés lors de changements de vitesse

de cyclage qui mettent en évidence la présence de réactions parasites. Afin d’étudier la

nature des réactions parasites, l’association de mesures en cellule de pression et de

spectrométrie de masse couplée au cyclage électrochimique ont montré que la réaction

de réduction de l’eau a lieu, à la fois pendant le cyclage et pendant les périodes de repos,

Résumé en français 245

i.e. pendant l’autodécharge, comme décrit par les Equation Résumé. 3, Equation

Résumé. 4 et Equation Résumé. 5.

Equation Résumé. 3 : Réactions (a) d’insertion des cations Li+ dans l’électrode négative LixM déchargée en parallèle de la (b) réduction directe de l’eau lors d’un cyclage continu.

(𝑎) 𝐿𝑖𝑥𝑀+ 𝑦 ∙ 𝐿𝑖+ + 𝑦 ∙ 𝑒− = 𝐿𝑖𝑥+𝑦𝑀

𝑤𝑖𝑡ℎ 0 ≤ 𝑥, 0 < 𝑦, 𝐿𝑖𝑥𝑀, insertion material

(𝑏) 2 ∙ 𝐻2𝑂 + 2 ∙ 𝑒− = 2 ∙ 𝐻𝑂− + 𝐻2

Equation Résumé. 4 : Demi-réaction redox du mécanisme d’autodécharge. (a) Désinsertion du matériau de négative chargé Li(x+y)M. (b) Réduction de l’eau.

(𝑎) 𝐿𝑖(𝑥+𝑦)𝑀 → 𝐿𝑖(𝑥+𝑦)−𝑧𝑀+ 𝑧 ∙ 𝐿𝑖+ + 𝑧 ∙ 𝑒−

0 ≤ 𝑥, 0 < 𝑦, 0 < 𝑧 ≤ 𝑦

(𝑏) 𝑧 ∙ 𝐻2𝑂 + 𝑧 ∙ 𝑒− → 𝑧 ∙ 𝐻𝑂− + 𝑧

2∙ 𝐻2

Equation Résumé. 5 : Réaction global du mécanisme d’autodécharge.

𝐿𝑖(𝑥+𝑦)𝑀 + 𝑧 ∙ 𝐻2𝑂 → 𝐿𝑖(𝑥+𝑦)−𝑧𝑀 + 𝑧 ∙ 𝐿𝑖+ + 𝑧 ∙ 𝐻𝑂− + 𝑧

2∙ 𝐻2

De plus, bien que le mécanisme d’autodécharge soit en partie réversible vis-à-vis du

lithium, la consommation d’eau est préjudiciable car irréversible. Enfin, l’utilisation de

20 m LiTFSI : 8 m LiBETI permet de compléter l’étude afin d’appréhender le

comportement général des électrolytes aqueux superconcentrés. De la même manière

que pour l’électrolyte basé sur 20 m LiTFSI, la réduction de l’eau a lieu pendant le cyclage

et les périodes de repos, mais avec une moindre intensité. A l’électrode positive, des

mesures de gaz ont montré que l’impact de l’oxydation de l’eau était limité. Le principal

obstacle des électrolytes aqueux superconcentrés provient donc de l’électrode négative

et de la stabilité de la SEI.

L’étude de la stabilité de la SEI native a été mené par des caractérisations

électrochimiques (voltamétrie cyclique et spectroscopie d’impédance électrochimique).

Les résultats montrent la passivation de la surface, mais la faible efficacité de celle-ci vis-

à-vis de la réduction de l’eau. Cette dernière est mise en valeur par des expériences de

calorimétrie à balayage différentiel qui atteste d’un décalage du pic de cristallisation de

l’électrolyte après des phases de cyclage continu et d’autodécharge témoignant de

l’augmentation de concentration. Par ailleurs, pour comprendre la différence de vitesses

de consommation d’eau lors des phases de cyclage et d’autodécharge, les énergies

246 Résumé en français

d’activation de la réduction directe de l’eau lors du cyclage et des phénomènes

d’autodécharge ayant lieu pendant les phases de repos ont été étudiées. Les énergies

d’activation trouvées sont semblables pour les phénomènes d’autodécharge et de

réduction de l’eau (≈ 20 kJ/mol), suggérant que l’autodécharge est gouvernée par la

réduction de l’eau plutôt que par la désinsertion du lithium. Pour résumer, l’impact

néfaste de la réduction de l’eau pendant l’autodécharge et le cyclage soulignent

l’instabilité de la SEI.

Aussi, le composé inorganique de fluorure de lithium (LiF) a été reporté comme l’un

des contributeurs majeurs de la SEI. L’évaluation de la solubilité de LiF dans différents

milieux tels que l’eau pure, 1 mol/kg LiTFSI, 20 mol/kg LiTFSI et 20 mol/kg LiTFSI : 8

mol/kg LiTFSI montre que, en considérant une épaisseur de LiF de ≈ 10 nm, semblable à

l’épaisseur d’une SEI native, LiF ne devrait pas être dissout dans les électrolytes aqueux

superconcentrés. Par conséquent, une couche artificielle de LiF déposée sur une

électrode de lithium métal, dénotée par la suite Li/LiF, a été utilisée pour imiter le

comportement d’une SEI native et évaluer l’efficacité d’une couche inorganique pour

prévenir de la réduction de l’eau.

La technique de microscopie électronique à balayage environnementale permet

d’observer la réactivité des échantillons de Li/LiF à une augmentation graduelle de

l’humidité dans la chambre du microscope (de 0 % RH à 90 % RH). Ces observations,

couplées à des mesures de chromatographie en phase gazeuse, démontrent que la

couche de LiF n’empêche pas l’eau d’accéder à l’interface de Li. Contrairement à ce qui

est observé lors de l’exposition d’un échantillon de Li/LiF à un électrolyte organique tel

que LP30 (1 mol/L LiPF6 in EC:DMC (1 :1)), l’exposition à un électrolyte aqueux

superconcentré ne permet pas d’auto-passiver l’échantillon. Cette absence d’auto-

passivation est confirmée par la réalisation de mesures similaires sur des échantillons

protégés par une couche de Al2O3. En conclusion, ces résultats montrent que, bien que

LiF ne se dissout pas dans les électrolytes aqueux superconcentrés, l’absence d’auto-

passivation ne prévient pas la réaction de réduction de l’eau qui a lieu à travers les

défauts de structures de la couche de LiF comme des fissures ou de la microporosité.

Pour surmonter ces obstacles, une étape de pré-imprégnation dans deux électrolytes

organiques, 1 mol/L LiTFSI : 2 mol/L LiFSI dans DOL:DME + 3 % LiNO3 ou 7 mol/L LiFSI

dans FEC, connus pour former des SEI riches en LiF dans leur couche interne et des

composés organiques dans leur couche externe, a été réalisée. Les mesures de

chromatographie en phase gazeuse montrent une diminution de la production

Résumé en français 247

d’hydrogène lors de l’exposition aux électrolytes aqueux superconcentrés des

échantillons pré-imprégnés. La réalisation de mesures par spectrométrie photo

électronique par rayons X (XPS) met en lumière le dépôt de composés issu de la

dégradation du sel provenant de la solution de pré-imprégnation. Par conséquent,

l’utilisation d’une couche artificielle de LiF couplée à la pré-imprégnation dans un

électrolyte organique démontrent le caractère essentiel des propriétés auto-

réparatrices de la SEI pour l’utilisation d’électrode négative dont le potentiel se situe en

deçà de la limite de stabilité de la fenêtre électrochimique.

Enfin, la réalisation d’une figure de mérite permet de comparer les performances des

batteries LIB basées sur les électrolytes aqueux superconcentrés avec celles des

batteries LIB commerciales ainsi que les performances des batteries aqueuses

rechargeables telles que les batteries Acide-plomb, Ni-Cd ou Ni-MH. L’utilisation d’un

électrolyte tel que 20 m LiTFSI limite la fenêtre de fonctionnement d’une batterie à

environ 2 V, afin d’éviter la contribution néfaste des réactions parasites. Ceci se fait au

détriment de la densité d’énergie et de l’énergie spécifique (limitées au niveau des

batteries type Ni-MH). De plus, alors que des solutions d’ingénierie ont été mis en place

dans les batteries aqueuses rechargeable afin de gérer les gaz générés pendant le

cyclage (O2 ou H2), la superconcentration peut causer la cristallisation rapide de

l’électrolyte dans le cas de cellule type « vented » ou « flooded » qui évacue

l’augmentation de pression sans compensation de l’eau consommée. Par ailleurs,

paradoxalement, peu d’oxygène est produit au niveau de l’électrode positive tandis que

la production d’hydrogène a lieu en parallèle de l’insertion des Li+ à la négative, ce qui

empêche l’implémentation du cycle de recombinaison de l’eau. En effet, dans les

cellules types « sealed » ou certaines cellules types « vented », la batterie est

dimensionnée afin que l’électrode positive soit limitante. En cas de surcharge, seule la

réaction parasite d’oxydation de l’eau doit avoir lieu, l’oxygène produit à la positive

diffuse alors dans l’électrolyte jusqu’à la négative où il est recombiné en eau, évitant à

la fois la consommation irréversible d’eau et la production d’hydrogène.

En dehors des considérations énergétiques, un inconvénient majeur des électrolytes

aqueux superconcentrés est leur haut point de fusion qui provoque la cristallisation

rapide de l’électrolyte et empêche généralement leur utilisation à des températures

inférieures à la température ambiante. L’introduction d’ions asymétriques permet

d’abaisser cette température et de diminuer la quantité d’eau dite libre, c’est-à-dire non

liée à un cation lithium en augmentant la solubilité. L’utilisation de co-solvant organique

modifie la sphère de solvatation des cations Li+, en réduisant la quantité d’eau dans la

248 Résumé en français

sphère de solvatation tout en promouvant la décomposition du co-solvant organique

lors de la formation de la SEI. Cependant, l’eau reste l’une des composantes de la

première sphère de solvatation des cations Li+ laissant penser que ces stratégies ne

permettent pas pour l’instant de résoudre le challenge cathodique, qui atteste de la

présence de molécule d’eau à l’interface de l’électrode négative à bas potentiel

favorisant leur réduction. Pour résumer, les électrolytes aqueux superconcentrés font

encore face à de nombreux freins qui empêchent, à l’heure actuelle, leur

commercialisation. Parmi les stratégies envisagées pour améliorer ces systèmes, l’une

d’elle repose sur l’intégration d’une couche polymère sur l’électrode négative pour à la

fois empêcher physiquement l’accès de l’eau à l’interface négative et participer à la

formation de la SEI par décomposition partielle ou totale de cette couche polymère.

Malgré la publication de cellules basées sur une électrode de graphite recouverte d’un

gel composé de 0.5 mol/L de LiTFSI dans un électrolyte basé sur le mélange HFE:DMC

(1,1,2,2-tetrafluoroethyl-2’,2’,2’-trifluoroethyl ether :dimethylcarbonate), nous n’avons

pour l’heure pas encore réussi à reproduire ces résultats afin de tester la viabilité de

cette solution.

Towards aqueous superconcentrated electrolytes and their use in Li-ion

battery

The development of superconcentrated aqueous electrolytes, namely Water-in-salt electrolytes (WiSE), from 2015 onwards has renewed the interest for aqueous-based Li-ion battery (LIB). Indeed, they were proposed to overcome issues related to safety and sustainability of common carbonate-based organic solvent while solving the poor performances of diluted aqueous electrolyte due to the narrow electrochemical stability window (ESW) of water (1.23 V). Such achievements are largely attributed to modification of the electrolyte structure upon increase in concentration that changes the physico-chemical properties and the interfacial reactivity. An inorganic LiF-based solid electrolyte interphase (SEI) was reported to be formed, opening the path for the use of low potential negative electrodes, further increasing the energy density of these batteries.

This work aims to provide answers regarding the viability of WiSE in LIB. By conducting a systematic study of the impact of superconcentration on battery performances as function of the operating conditions, we demonstrate that the SEI is not able to prevent water reduction following the hydrogen evolution reaction (HER), neither during cycling nor during resting period, i.e. self-discharge. Indeed, the rates for water consumption calculated during cycling and resting period are found within the same order of magnitude, highlighting the SEI limitation to prevent water reduction although the surface is passivated. Determining the activation energies for HER during cycling and self-discharge, we suggest that self-discharge is more likely driven by water reduction than Li+ deintercalation. Eventually, LiF solubility measurements, gas chromatography tests and environmental scanning electron microscopy suggest that SEI instability is related to structural defects that cannot be self-passivated in WiSE. A presoaking step in organic electrolyte of an artificial Li/LiF layer reduces water consumption and thus confirms the need for the SEI to self-repair.

Keywords: aqueous Li-ion batteries, Water-in-salt, SEI stability, superconcentrated electrolyte

Vers des électrolytes aqueux superconcentrés et leur application pour

une utilisation batterie Li-ion

Depuis 2015, le développement des électrolytes aqueux superconcentrés, dénommés « Water-in-salt electrolytes » (WiSE), a suscité un regain d’intérêt pour les batteries Li-ion (LIB) aqueuses. Les WiSE proposent une alternative aux électrolytes organiques commerciaux qui posent des problèmes de sécurité et de durabilité, tout en résolvant les faibles performances des électrolytes aqueux dilués limitées par l’étroitesse de la fenêtre électrochimique (1 .23 V). En effet, la superconcentration influe sur les propriétés physico-chimiques et la réactivité interfaciale. La formation d’une interphase solide/électrolyte inorganique (SEI) riche en fluorure de lithium (LiF) ouvre la voie à l’utilisation d’électrode négative à bas potentiel et donc à l’augmentation de la densité d’énergie de ces batteries.

Cette thèse étudie la viabilité des électrolytes WiSE dans les LIB. Grâce à la mise en place d’une étude systématique, l’impact de la superconcentration sur les performances des batteries en fonction des conditions d’opération montre que la SEI formée ne prévient pas de la réduction de l’eau, appelée réaction d’évolution de l’hydrogène (HER) ni pendant le cyclage ni pendant les périodes de repos, i.e. l’autodécharge. L’évaluation des vitesses de consommation de l’eau souligne les limites des propriétés protectrices de la SEI malgré la passivation de l’interface. Par ailleurs, la détermination des énergies d’activation de la HER directe, ayant lieu pendant le cyclage, et du phénomène d’autodécharge suggère que l’autodécharge est gouvernée par la HER. Enfin, l’évaluation de la solubilité de LiF dans les WiSE, des observations au microscope environnemental à balayage électronique et des mesures de chromatographie en phase gaz suggèrent que l’instabilité de la SEI est d’avantage reliée à des défauts microstructuraux qui ne peuvent pas être comblés dû à l’absence d’auto-passivation de l’interface. Une étape de pré-imprégnation dans un électrolyte organique réduit la consommation d’eau, confirmant la nécessité de propriétés d’autoréparation de la SEI.

Mots-Clés : Batteries Li-ion aqueuses, Water-in-salt, stabilité de la SEI, électrolytes superconcentrés